hn_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Table of Contents"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Import Modules"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's important modules for analysis, visualizations and modeling."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from pymongo import MongoClient\n",
    "import time\n",
    "import re\n",
    "import html\n",
    "from sklearn.decomposition import TruncatedSVD, LatentDirichletAllocation, NMF\n",
    "from sklearn.preprocessing import Normalizer\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import nltk\n",
    "from nltk import word_tokenize\n",
    "from nltk.corpus import stopwords\n",
    "from nltk.stem import WordNetLemmatizer \n",
    "from sklearn.feature_extraction.text import TfidfVectorizer, CountVectorizer\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.cluster import KMeans, MiniBatchKMeans, DBSCAN\n",
    "from sklearn.metrics import silhouette_score, silhouette_samples\n",
    "from gensim import corpora, models, similarities, matutils\n",
    "import logging"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Fetch Data from Database"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I previously made a query to my MongoDB database to fetch all the Hacker News documents recorded. Once I had enough data, I pickled a dataframe to use for analysis. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# client = MongoClient(\"mongodb:/{ip_address_for_mongo_server}/hacker_news_database\")\n",
    "# mydb = client.hacker_news_database  # names a database\n",
    "# collection = mydb.hacker_news_collection  # refers to my collection\n",
    "# data = pd.DataFrame(list(collection.find()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Turn our observations (Hacker News posts/comments) into rows. Each key from an API request is seen here as a column with its corresponding value for each observation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "data = pd.read_pickle('three_hundred_K_hn_posts_df')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>_id</th>\n",
       "      <th>by</th>\n",
       "      <th>dead</th>\n",
       "      <th>deleted</th>\n",
       "      <th>descendants</th>\n",
       "      <th>kids</th>\n",
       "      <th>parent</th>\n",
       "      <th>parts</th>\n",
       "      <th>poll</th>\n",
       "      <th>score</th>\n",
       "      <th>text</th>\n",
       "      <th>time</th>\n",
       "      <th>title</th>\n",
       "      <th>type</th>\n",
       "      <th>url</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>14170376</td>\n",
       "      <td>dragonwriter</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[14170821]</td>\n",
       "      <td>14169290.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Well, there&amp;#x27;s an interesting, tangentiall...</td>\n",
       "      <td>1492819269</td>\n",
       "      <td>NaN</td>\n",
       "      <td>comment</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>14170375</td>\n",
       "      <td>uncensored</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>14170058.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Substansively? Tons of replies are not substan...</td>\n",
       "      <td>1492819230</td>\n",
       "      <td>NaN</td>\n",
       "      <td>comment</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>14170374</td>\n",
       "      <td>supernumerary</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[14170677, 14170866, 14173394]</td>\n",
       "      <td>14168157.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Regarding chicago this: \\n&lt;a href=\"https:&amp;#x2F...</td>\n",
       "      <td>1492819229</td>\n",
       "      <td>NaN</td>\n",
       "      <td>comment</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>14170373</td>\n",
       "      <td>bobsil1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>14169489.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Parking lots we don&amp;#x27;t need after autonomo...</td>\n",
       "      <td>1492819210</td>\n",
       "      <td>NaN</td>\n",
       "      <td>comment</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>14170372</td>\n",
       "      <td>thecrazyone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>14168900.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>It&amp;#x27;s good to have this as a general remin...</td>\n",
       "      <td>1492819202</td>\n",
       "      <td>NaN</td>\n",
       "      <td>comment</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        _id             by dead deleted  descendants  \\\n",
       "0  14170376   dragonwriter  NaN     NaN          NaN   \n",
       "1  14170375     uncensored  NaN     NaN          NaN   \n",
       "2  14170374  supernumerary  NaN     NaN          NaN   \n",
       "3  14170373        bobsil1  NaN     NaN          NaN   \n",
       "4  14170372    thecrazyone  NaN     NaN          NaN   \n",
       "\n",
       "                             kids      parent parts  poll  score  \\\n",
       "0                      [14170821]  14169290.0   NaN   NaN    NaN   \n",
       "1                             NaN  14170058.0   NaN   NaN    NaN   \n",
       "2  [14170677, 14170866, 14173394]  14168157.0   NaN   NaN    NaN   \n",
       "3                             NaN  14169489.0   NaN   NaN    NaN   \n",
       "4                             NaN  14168900.0   NaN   NaN    NaN   \n",
       "\n",
       "                                                text        time title  \\\n",
       "0  Well, there&#x27;s an interesting, tangentiall...  1492819269   NaN   \n",
       "1  Substansively? Tons of replies are not substan...  1492819230   NaN   \n",
       "2  Regarding chicago this: \\n<a href=\"https:&#x2F...  1492819229   NaN   \n",
       "3  Parking lots we don&#x27;t need after autonomo...  1492819210   NaN   \n",
       "4  It&#x27;s good to have this as a general remin...  1492819202   NaN   \n",
       "\n",
       "      type  url  \n",
       "0  comment  NaN  \n",
       "1  comment  NaN  \n",
       "2  comment  NaN  \n",
       "3  comment  NaN  \n",
       "4  comment  NaN  "
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "296,745 documents in our MongoDB collection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "296745"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(data.index)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Clean Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 296745 entries, 0 to 296744\n",
      "Data columns (total 15 columns):\n",
      "_id            296745 non-null int64\n",
      "by             287920 non-null object\n",
      "dead           14071 non-null object\n",
      "deleted        8824 non-null object\n",
      "descendants    32306 non-null float64\n",
      "kids           125967 non-null object\n",
      "parent         253433 non-null float64\n",
      "parts          4 non-null object\n",
      "poll           11 non-null float64\n",
      "score          41742 non-null float64\n",
      "text           248388 non-null object\n",
      "time           296745 non-null int64\n",
      "title          41731 non-null object\n",
      "type           296745 non-null object\n",
      "url            39260 non-null object\n",
      "dtypes: float64(4), int64(2), object(9)\n",
      "memory usage: 34.0+ MB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have 296,745 observations (Hacker News posts/comments)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RangeIndex(start=0, stop=296745, step=1)"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.index"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I noticed the first few Hacker News posts contained characters not seen on the site. There were HTML tags and escape characters. We ought to remove them so they don't skew our natural language processing insights.\n",
    "\n",
    "I noticed some text values appear as NaN values that are floats too."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df = data.dropna(subset=['text'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def remove_html(hn_comment):\n",
    "    \"\"\"\n",
    "    Convert all numeric character references like &gt and &#62 into corresponding unicode characters. \n",
    "    Remove HTML tags for paragraphs and italics.\n",
    "    Source: https://docs.python.org/3/library/html.html\n",
    "    \"\"\"\n",
    "    hn_comment = html.unescape(hn_comment)\n",
    "    return hn_comment"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Apply remove_html function to text column in our dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/dan/anaconda/lib/python3.6/site-packages/pandas/core/generic.py:2773: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  self[name] = value\n"
     ]
    }
   ],
   "source": [
    "df.text = df.text.apply(remove_html)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>_id</th>\n",
       "      <th>by</th>\n",
       "      <th>dead</th>\n",
       "      <th>deleted</th>\n",
       "      <th>descendants</th>\n",
       "      <th>kids</th>\n",
       "      <th>parent</th>\n",
       "      <th>parts</th>\n",
       "      <th>poll</th>\n",
       "      <th>score</th>\n",
       "      <th>text</th>\n",
       "      <th>time</th>\n",
       "      <th>title</th>\n",
       "      <th>type</th>\n",
       "      <th>url</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>14170376</td>\n",
       "      <td>dragonwriter</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[14170821]</td>\n",
       "      <td>14169290.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Well, there's an interesting, tangentially rel...</td>\n",
       "      <td>1492819269</td>\n",
       "      <td>NaN</td>\n",
       "      <td>comment</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>14170375</td>\n",
       "      <td>uncensored</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>14170058.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Substansively? Tons of replies are not substan...</td>\n",
       "      <td>1492819230</td>\n",
       "      <td>NaN</td>\n",
       "      <td>comment</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>14170374</td>\n",
       "      <td>supernumerary</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[14170677, 14170866, 14173394]</td>\n",
       "      <td>14168157.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Regarding chicago this: \\n&lt;a href=\"https://en....</td>\n",
       "      <td>1492819229</td>\n",
       "      <td>NaN</td>\n",
       "      <td>comment</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>14170373</td>\n",
       "      <td>bobsil1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>14169489.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Parking lots we don't need after autonomous cars.</td>\n",
       "      <td>1492819210</td>\n",
       "      <td>NaN</td>\n",
       "      <td>comment</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>14170372</td>\n",
       "      <td>thecrazyone</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>14168900.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>It's good to have this as a general reminder: ...</td>\n",
       "      <td>1492819202</td>\n",
       "      <td>NaN</td>\n",
       "      <td>comment</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        _id             by dead deleted  descendants  \\\n",
       "0  14170376   dragonwriter  NaN     NaN          NaN   \n",
       "1  14170375     uncensored  NaN     NaN          NaN   \n",
       "2  14170374  supernumerary  NaN     NaN          NaN   \n",
       "3  14170373        bobsil1  NaN     NaN          NaN   \n",
       "4  14170372    thecrazyone  NaN     NaN          NaN   \n",
       "\n",
       "                             kids      parent parts  poll  score  \\\n",
       "0                      [14170821]  14169290.0   NaN   NaN    NaN   \n",
       "1                             NaN  14170058.0   NaN   NaN    NaN   \n",
       "2  [14170677, 14170866, 14173394]  14168157.0   NaN   NaN    NaN   \n",
       "3                             NaN  14169489.0   NaN   NaN    NaN   \n",
       "4                             NaN  14168900.0   NaN   NaN    NaN   \n",
       "\n",
       "                                                text        time title  \\\n",
       "0  Well, there's an interesting, tangentially rel...  1492819269   NaN   \n",
       "1  Substansively? Tons of replies are not substan...  1492819230   NaN   \n",
       "2  Regarding chicago this: \\n<a href=\"https://en....  1492819229   NaN   \n",
       "3  Parking lots we don't need after autonomous cars.  1492819210   NaN   \n",
       "4  It's good to have this as a general reminder: ...  1492819202   NaN   \n",
       "\n",
       "      type  url  \n",
       "0  comment  NaN  \n",
       "1  comment  NaN  \n",
       "2  comment  NaN  \n",
       "3  comment  NaN  \n",
       "4  comment  NaN  "
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We still have HTML tags and \\n in our comments. Let's use regex to remove them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/dan/anaconda/lib/python3.6/site-packages/ipykernel/__main__.py:4: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n"
     ]
    }
   ],
   "source": [
    "# wanted to pass parameter inplace=True but kept on getting an error; so making a copy instead\n",
    "pattern = r\"<.*?>\"\n",
    "replace = \" \"\n",
    "df['text'] = df['text'].str.replace(pattern, replace)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/dan/anaconda/lib/python3.6/site-packages/ipykernel/__main__.py:3: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  app.launch_new_instance()\n"
     ]
    }
   ],
   "source": [
    "pattern = r\"\\n\"\n",
    "replace = \" \"\n",
    "df['text'] = df['text'].str.replace(pattern, replace)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create a dataframe just of the text comment values from HN posts. We'll start with NLP analysis of just these."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df_type_comments = df.loc[df.type == 'comment']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create Vector Space from Text"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We want to derive meaning from the natural language of Hacker News comments. We must map the langauge to a vector space for analysis."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[Stemming](https://en.wikipedia.org/wiki/Stemming) allows us to reduce derived words to their word stem - aka base or root form.\n",
    "\n",
    "\"Amazon's catalog contains\" -> \"Amazon catalog contain\"\n",
    "\n",
    "Notice how the possessive form of Amazon was removed and the \"s\" ending in the word \"contains.\"\n",
    "\n",
    "[Lemmatization](https://en.wikipedia.org/wiki/Lemmatisation) is often viewed as the more powerful form of stemming. Lemming understands vocabulary and morphological analysis. Essentially, it removes inflectional endings to return the word to its base or dictionary form of a word, defined as the lemma.\n",
    "\n",
    "[Sklearn doc on using Nltk Lemmatizer](http://scikit-learn.org/stable/modules/feature_extraction.html#customizing-the-vectorizer-classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class LemmaTokenizer(object):\n",
    "    \"\"\"\n",
    "    References NLTK Lemmatization class and functionality to modify words in corpus\n",
    "    \"\"\"\n",
    "    def __init__(self):\n",
    "        self.wnl = WordNetLemmatizer()  # a call to NLTK\n",
    "    def __call__(self, doc):\n",
    "        return [self.wnl.lemmatize(t) for t in word_tokenize(doc)]  # apply lematize for each word"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The LemmaTokenizer may have caused some issue "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "count_vectorizer = CountVectorizer(min_df=10,\n",
    "                                   stop_words='english', \n",
    "                                   token_pattern='[a-z][a-z]+', \n",
    "                                   ngram_range=(1,2))\n",
    "count_vect_matrix = count_vectorizer.fit_transform(df_type_comments.text)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "TfidfVectorizer counts words in each document, and weights the counts relative to how those words appear across all documents.\n",
    "\n",
    "- min_df=10 only includes words we've seen at least 10 times\n",
    "- stop_words='english' exludes words that have very little semantic value (such as 'a', 'an', etc)\n",
    "- token_pattern='[a-z][a-z]+' makes sure we only include alphabetical-based words, not numbers.\n",
    "- ngram_range=(1,2) sets lower and upper boundary of n-values; for features we'll have words and 2-word phrases (aka bigrams)\n",
    "\n",
    "We call fit_transform() on our data to transform our data into our new features into a dimensional space that corresponds to our parameters with Tfidf\n",
    "\n",
    "[Sklearn documentation on TfidfVectorizer](http://scikit-learn.org/stable/modules/generated/sklearn.feature_extraction.text.TfidfVectorizer.html)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "vectorizer = TfidfVectorizer(min_df = 10,\n",
    "                             stop_words = 'english',\n",
    "                             token_pattern = '[a-z][a-z]+',\n",
    "                             ngram_range = (1,2))\n",
    "X = vectorizer.fit_transform(df_type_comments.text)  # returns a Scipy sparse matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have 83,175 features after our Tfidf vectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "83175"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(vectorizer.get_feature_names())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We want to reduce the number of features under consideration, while minimizing information loss in our data. We'll use dimensionality reduction to reduce the number of features, but create new ones that help explain the most variance in our data. The first principal component has the largest possible variance and each succeeding component in turn has the higest variance possible and uncorrelated with the preceding components. \n",
    "\n",
    "We use Truncated SVD (singular value decomposition through latent samentic analysis) because it works with sparse matrices efficiently. SVD works on Tfidf matrixes as well. \n",
    "\n",
    "[Sklearn documentation on TruncatedSVD](http://scikit-learn.org/stable/modules/generated/sklearn.decomposition.TruncatedSVD.html)\n",
    "\n",
    "- n_components=40 examines first 40 principal components (latent semantic features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<246175x83175 sparse matrix of type '<class 'numpy.float64'>'\n",
       "\twith 7508820 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I want to see what our matrix decomposition looks like under the hood."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.utils.extmath import randomized_svd\n",
    "\n",
    "U, Sigma, VT = randomized_svd(X, \n",
    "                              n_components=15,\n",
    "                              n_iter=5,\n",
    "                              random_state=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.00184738,  0.00022025,  0.00050301, ..., -0.00115457,\n",
       "        -0.00234192,  0.00020745],\n",
       "       [ 0.00083195, -0.00038486, -0.00026842, ...,  0.00496335,\n",
       "        -0.00031413, -0.00135294],\n",
       "       [ 0.00342035,  0.00406801,  0.00965432, ...,  0.00115994,\n",
       "         0.00016543,  0.00089364],\n",
       "       ..., \n",
       "       [ 0.00152403, -0.00050143, -0.0001921 , ..., -0.00035186,\n",
       "        -0.00014684, -0.00037306],\n",
       "       [ 0.0016524 , -0.00054557, -0.00039051, ..., -0.00282791,\n",
       "        -0.00438656,  0.00125054],\n",
       "       [ 0.00093931, -0.00042672, -0.00022989, ..., -0.00039535,\n",
       "         0.00016842, -0.00048218]])"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "U"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 38.19621221,  31.0574047 ,  25.62477408,  23.24427463,\n",
       "        18.8195793 ,  18.30812046,  17.51070779,  16.95583995,\n",
       "        16.66351314,  16.58736918,  15.52554755,  15.23777116,\n",
       "        15.154738  ,  15.01380169,  14.83673161])"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Sigma"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[  1.06167549e-03,   9.15462554e-04,   1.78764062e-04, ...,\n",
       "          4.45853518e-04,   2.24947076e-04,   1.20783508e-04],\n",
       "       [  3.43585846e-04,  -3.70381319e-04,  -5.93248948e-05, ...,\n",
       "          2.08787206e-05,  -2.85200863e-05,  -3.83400741e-05],\n",
       "       [  7.41168100e-04,  -1.59251858e-04,  -1.51800972e-05, ...,\n",
       "          1.43440778e-04,   5.37893230e-05,   6.21964131e-06],\n",
       "       ..., \n",
       "       [ -1.54683210e-04,   9.15389737e-06,  -1.42116875e-05, ...,\n",
       "         -2.64270490e-04,   6.70619141e-05,   7.44102927e-05],\n",
       "       [ -6.42444255e-04,  -5.23082239e-05,   1.72739159e-05, ...,\n",
       "         -8.34697983e-04,   6.50970536e-05,   7.61175970e-05],\n",
       "       [  2.78141165e-04,   3.95328078e-04,  -1.11011067e-04, ...,\n",
       "         -1.43343031e-04,   1.08659656e-04,  -2.37087356e-05]])"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "VT"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "number_of_principle_components = 40\n",
    "svd = TruncatedSVD(n_components=number_of_principle_components)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Normalization helps scale individual samples to have unit norm. This is useful as we plan to use the dot product of matrices to quantify the similarity of pairs of samples. \n",
    "\n",
    "Normalization is important if you have different length documents (ex. a two word document versus a 200 word document). https://stats.stackexchange.com/questions/69157/why-do-we-need-to-normalize-data-before-analysis\n",
    "\n",
    "\n",
    "[Sklearn documentation on Normalizer](http://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.Normalizer.html#sklearn.preprocessing.Normalizer)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "normalizer = Normalizer(copy=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Think of this pipeline as just a grouping for a procedure of steps.\n",
    "\n",
    "[Sklearn documentation on make_pipeline](http://scikit-learn.org/stable/modules/generated/sklearn.pipeline.make_pipeline.html)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "lsa = make_pipeline(svd, normalizer)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We call fit_transform becaause we're transforming our data to be returned in a new matrix with new (latent semantic) features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "X = lsa.fit_transform(X)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These values below represent our new linear coefficients that came from the transformation from our original space into the PCA space.\n",
    "\n",
    "Each of these eigenvalues for each observation corresponds to the amount of variance explained by that component."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As expected, we now have observations of all our HN comments, and just 40 new components."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(246175, 40)"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[  1.06167495e-03,   9.15463427e-04,   1.78763874e-04, ...,\n",
       "          4.45855519e-04,   2.24947278e-04,   1.20783831e-04],\n",
       "       [  3.43556669e-04,  -3.70419434e-04,  -5.93272842e-05, ...,\n",
       "          2.09248317e-05,  -2.85367001e-05,  -3.83561974e-05],\n",
       "       [  7.40702966e-04,  -1.59168724e-04,  -1.50517673e-05, ...,\n",
       "          1.42989358e-04,   5.37577816e-05,   6.19336690e-06],\n",
       "       [ -1.23573847e-03,  -5.59917603e-05,  -4.63726918e-05, ...,\n",
       "         -3.02805762e-04,  -1.10066801e-04,  -3.28637341e-05],\n",
       "       [  1.09214495e-04,  -1.38763584e-05,   5.49586122e-06, ...,\n",
       "         -8.23294474e-05,   3.29855486e-05,  -7.18665130e-06]])"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eigvecs = svd.components_\n",
    "eigvecs[0:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The values below detail the variance of the principle components from one another. We have a tradeoff; we want fewer principle components in order to make a simpler model, but we don't want to oversimplify. There's diminishing margins of returns for additional principle components. We're looking for the \"elbow\" below - a break between useful principle components and residual noise. The ideal principle components is probably around 9.\n",
    "\n",
    "There's an issue here - in which our first principal component explains less variance than the next few principal components. It may be an issue with Sklearn. I'll investigate more.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.00135986,  0.00390105,  0.00268937,  0.00221295,  0.00144631,\n",
       "        0.00137348,  0.00125472,  0.00118229,  0.00114262,  0.00112774,\n",
       "        0.00100116,  0.00096526,  0.00095952,  0.00094668,  0.00093267,\n",
       "        0.00091068,  0.00090357,  0.00087436,  0.00086438,  0.00084561,\n",
       "        0.0008383 ,  0.00080671,  0.00078518,  0.00077217,  0.0007638 ,\n",
       "        0.00074437,  0.00073954,  0.00072643,  0.0007146 ,  0.00069311,\n",
       "        0.00067354,  0.00067085,  0.00066333,  0.00065203,  0.00063871,\n",
       "        0.00063761,  0.00062904,  0.00062326,  0.000619  ,  0.00060209])"
      ]
     },
     "execution_count": 86,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svd.explained_variance_ratio_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.00134868,  0.00386898,  0.00266726,  0.00219476,  0.00143442,\n",
       "        0.00136219,  0.0012444 ,  0.00117257,  0.00113322,  0.00111847,\n",
       "        0.00099293,  0.00095732,  0.00095163,  0.0009389 ,  0.000925  ,\n",
       "        0.00090319,  0.00089614,  0.00086718,  0.00085727,  0.00083866,\n",
       "        0.00083141,  0.00080008,  0.00077872,  0.00076582,  0.00075752,\n",
       "        0.00073825,  0.00073346,  0.00072046,  0.00070873,  0.00068741,\n",
       "        0.000668  ,  0.00066534,  0.00065787,  0.00064667,  0.00063345,\n",
       "        0.00063237,  0.00062387,  0.00061814,  0.00061391,  0.00059714])"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svd.explained_variance_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.13598576,  0.39010486,  0.26893719,  0.22129542,  0.14463096,\n",
       "        0.13734834,  0.12547183,  0.11822904,  0.11426152,  0.11277441,\n",
       "        0.10011631,  0.09652557,  0.09595175,  0.09466846,  0.09326654,\n",
       "        0.09106778,  0.09035656,  0.08743638,  0.08643779,  0.08456071,\n",
       "        0.08382978,  0.08067089,  0.07851789,  0.07721712,  0.07637973,\n",
       "        0.07443727,  0.07395378,  0.07264284,  0.07146044,  0.06931095,\n",
       "        0.06735403,  0.06708509,  0.06633265,  0.06520337,  0.06387051,\n",
       "        0.06376131,  0.06290418,  0.06232591,  0.06190032,  0.06020919])"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "percent_explained_variance_features = svd.explained_variance_ratio_*100\n",
    "percent_explained_variance_features"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With these new 40 features, we can now explain 5.4% of the variance in total between these 40 features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4.0887944234197509"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum_explained_variance = svd.explained_variance_ratio_.sum()*100\n",
    "sum_explained_variance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x17ced8160>"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEWCAYAAAB1xKBvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nThhmZpbFCcPMzLL8D+guq9E7iU42AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x140f8b400>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.bar(list(range(1, number_of_principle_components+1)), percent_explained_variance_features)\n",
    "plt.xlabel('Principle Component')\n",
    "plt.ylabel('Percent Explained')\n",
    "plt.title('Explained Variance')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.13598576,  0.52609061,  0.7950278 ,  1.01632323,  1.16095418,\n",
       "        1.29830253,  1.42377435,  1.54200339,  1.65626492,  1.76903933,\n",
       "        1.86915564,  1.96568121,  2.06163297,  2.15630142,  2.24956796,\n",
       "        2.34063574,  2.4309923 ,  2.51842868,  2.60486646,  2.68942717,\n",
       "        2.77325695,  2.85392784,  2.93244573,  3.00966285,  3.08604258,\n",
       "        3.16047985,  3.23443364,  3.30707647,  3.37853691,  3.44784787,\n",
       "        3.5152019 ,  3.58228698,  3.64861963,  3.713823  ,  3.77769351,\n",
       "        3.84145482,  3.904359  ,  3.96668491,  4.02858523,  4.08879442])"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cum_sum_explained_variance = np.cumsum(percent_explained_variance_features)\n",
    "cum_sum_explained_variance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x16ae47cc0>"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nmdU4KZiZWY2TgpmZ1TgpmJlZzf8HA5XqZpcheSYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1626eedd8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.bar(list(range(1, number_of_principle_components+1)), cum_sum_explained_variance)\n",
    "plt.xlabel('Principal Component')\n",
    "plt.ylabel('Cumulative Explained Variance')\n",
    "plt.title(\"Total Components' Explained Variance\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our SVD (of 40 latent features) only explains ~4% of variance of the data. There's a huge amount of variance over our millions of Hacker News comments because topics range from business, to programming, to science and nearly everything else. Therefore, ~4% is fairly reasonable. If we were using data for fairly homogenous documents, we'd expect the sum of our components to explain much closer to 100% variance."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Clustering"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The Hacker News comments data set doesn't have labels. It's an instructured data set and we're trying to make meaning/structure out of it by grouping documents based on their features and features' values (Tfidf weights) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Intertia is the sum of square distances in each cluster. A low inertia value = high density of points of observations around the center of its cluster. More clusters likely equates to a lower inertia value for our KMeans model. If we fit our model multiple times, we'd get slightly different inertia values. We want to minimize our inertia values for a given number of clusters.\n",
    "\n",
    "We want to find a good estimate for the number of clusters we should utilize for a KMeans model. To do this, let's plot our inertia values over several values of clusters. We're looking for the elbow in the plot - the point of diminishing inertia values for each additional cluster.\n",
    "\n",
    "However, upon implementing the code below, I had massive fluctuations in inertia with each incremental number of clusters. The problem with using inertia is that it doesn't penalize for additional number of clusters. Generally, we want fewer clusters in order to easily make understandable groups."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# inertia_values = []\n",
    "# range_of_K_values = range(1, 25)\n",
    "# for num_clusters in range_of_K_values:\n",
    "#     kmeans = KMeans(n_clusters=number_of_clusters, init='k-means++', max_iter=100, n_init=1, n_jobs=-1)\n",
    "#     kmeans.fit(X)\n",
    "#     inertia_values.append(kmeans.inertia_)\n",
    "\n",
    "# inertia_values\n",
    "\n",
    "# # graph inertia over number of clusters\n",
    "# plt.plot(list(range_of_K_values), inertia_values)\n",
    "# plt.xlabel('Number of Clusters')\n",
    "# plt.ylabel('Kmeans Inertia')\n",
    "# plt.title('Cluster Density')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Instead, I tested out various number of clusters - from 8 to 25. I found 20 clusters to be optimal. At 20 clusters, I noticed clusters that revealed words across documents that were within the scope of documents on Hacker News - for words around programming languages, tech companies, code, etc. \n",
    "\n",
    "For example, one cluster may talk about \"app, microsoft, linux\" while another \"free, open, open source\" and another \"money, market, uber\". \n",
    "\n",
    "I tried out many combinations of clusters smaller than 20, but couldn't glean major differences between clusters with generic words such as \"work, time, great\". These words could be associated with nearly any forum site; however, Hacker News is generally more tech/science/business oriented so I cared more about words in these clusters."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "‘k-means++’ selects initial cluster centers for KMeans clustering in a smart way to speed up convergence. This is especially helpful as we have high dimensionality data.\n",
    "\n",
    "[sklearn documentation on KMeans](http://scikit-learn.org/stable/modules/generated/sklearn.cluster.KMeans.html#sklearn.cluster.KMeans)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "number_of_clusters = 20\n",
    "km = KMeans(n_clusters=number_of_clusters, init='k-means++', max_iter=100, n_init=1, n_jobs=-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I also tried getting the silhouette score (below). That can be considered a good metric to help identify your optimal number of clusters. However, I couldn't get it to work as my Kernal would die everytime. I believe 200K samples is outside the performance capable of this algorithm to get silhouette scores.\n",
    "\n",
    "[sklearn documentation on silhouette score](http://scikit-learn.org/stable/auto_examples/cluster/plot_kmeans_silhouette_analysis.html#sphx-glr-auto-examples-cluster-plot-kmeans-silhouette-analysis-py)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# X.shape\n",
    "# predicted_labels = km.fit_predict(X)\n",
    "# predicted_labels.shape\n",
    "# predicted_labels[0:100]\n",
    "# set(predicted_labels)\n",
    "# silhouette_avg = silhouette_score(X, predicted_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We already did fit_transform above with SVD to get our new matrix shape that contains our 40 latent samentic features. Here, we can simply fit our data to our KMeans model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KMeans(algorithm='auto', copy_x=True, init='k-means++', max_iter=100,\n",
       "    n_clusters=20, n_init=1, n_jobs=-1, precompute_distances='auto',\n",
       "    random_state=None, tol=0.0001, verbose=0)"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "km.fit(X)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "km.cluster_centers_ takes coordinates of cluster centers\n",
    "\n",
    "svd.inverse_transform transforms the dataset of cluster centers back to the corresponding dataset before SVD truncatation (which is after Tfidf vectorizing). So we have the Tfidf weights as values in our matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have 20 clusters, and 40 latent semantic features. So each cluster has coordinates in 40-dimensional space."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(20, 40)"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "km.cluster_centers_.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.39620947, -0.119902  , -0.05627003,  0.014505  ,  0.03419883,\n",
       "        0.02918329,  0.04877089,  0.07289593, -0.07694694, -0.02672751,\n",
       "       -0.16119874, -0.01501859, -0.00937057, -0.07997836, -0.05260556,\n",
       "        0.24681106,  0.03055058,  0.03311661, -0.05080965,  0.03037132,\n",
       "       -0.03740265, -0.148166  ,  0.05908974,  0.40807326, -0.09541083,\n",
       "       -0.03822647, -0.15304457,  0.01979978, -0.03672623,  0.04611517,\n",
       "       -0.00622793, -0.00368075,  0.04702796,  0.01673997,  0.0061776 ,\n",
       "       -0.02864281, -0.03912132,  0.01664593, -0.02237636, -0.02151007])"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "km.cluster_centers_[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ".argsort returns the indices that would sort the array from its smallest values to largest\n",
    "\n",
    "[numpy documentation on argsort](https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0ahUKEwiWj6fNhJbUAhUGRiYKHVeWA0sQFggoMAA&url=https%3A%2F%2Fdocs.scipy.org%2Fdoc%2Fnumpy%2Freference%2Fgenerated%2Fnumpy.argsort.html&usg=AFQjCNEIZw-eAurbq7M8XJpjdmaPTFdxtw&sig2=192Af96YCi6cKgajcMVOCQ)\n",
    "\n",
    "[:, ::-1] gets all values from list, starting at highest index\n",
    "\n",
    "So order_centroids returns us rows as cluster values - with each value being the largest Tfidf values for that cluster"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "original_space_centroids = svd.inverse_transform(km.cluster_centers_)\n",
    "order_centroids = original_space_centroids.argsort()[:, ::-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[77403, 63074, 77627, ..., 31738, 11357, 28664],\n",
       "       [80739, 35330, 81294, ..., 66131, 74608, 37681],\n",
       "       [31609, 31688, 11069, ..., 83059, 79039, 83056],\n",
       "       ..., \n",
       "       [58348, 35736, 39161, ..., 11357, 30996, 40386],\n",
       "       [75796, 35736, 76336, ..., 82384, 11357, 83059],\n",
       "       [79118,  5760, 28664, ..., 31820, 74608, 69761]])"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "order_centroids[0:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(20, 83175)"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "order_centroids.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We want to get the feature names after the Tfidf vectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "terms = vectorizer.get_feature_names()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top terms per cluster by Tfidf ranking:\n",
      "Cluster 0: \n",
      " ve\n",
      " seen\n",
      " ve seen\n",
      " got\n",
      " used\n",
      " heard\n",
      " ve heard\n",
      " years\n",
      " just\n",
      " read\n",
      " did\n",
      " actually\n",
      " things\n",
      " ve got\n",
      " like\n",
      " people\n",
      " lot\n",
      " don\n",
      " using\n",
      " probably\n",
      " \n",
      " \n",
      "Cluster 1: \n",
      " work\n",
      " job\n",
      " working\n",
      " day\n",
      " company\n",
      " doesn\n",
      " time\n",
      " people\n",
      " just\n",
      " hours\n",
      " like\n",
      " doing\n",
      " need\n",
      " hard\n",
      " way\n",
      " lot\n",
      " home\n",
      " does\n",
      " think\n",
      " pay\n",
      " \n",
      " \n",
      "Cluster 2: \n",
      " http\n",
      " http www\n",
      " com\n",
      " www\n",
      " org\n",
      " news\n",
      " html\n",
      " article\n",
      " like\n",
      " uk\n",
      " net\n",
      " blog\n",
      " data\n",
      " app\n",
      " just\n",
      " way\n",
      " did\n",
      " https www\n",
      " doesn\n",
      " people\n",
      " \n",
      " \n",
      "Cluster 3: \n",
      " money\n",
      " uber\n",
      " company\n",
      " pay\n",
      " market\n",
      " companies\n",
      " business\n",
      " tax\n",
      " new\n",
      " cost\n",
      " year\n",
      " make\n",
      " price\n",
      " government\n",
      " free\n",
      " service\n",
      " going\n",
      " car\n",
      " high\n",
      " product\n",
      " \n",
      " \n",
      "Cluster 4: \n",
      " does\n",
      " google\n",
      " code\n",
      " thanks\n",
      " yes\n",
      " data\n",
      " years\n",
      " source\n",
      " did\n",
      " article\n",
      " language\n",
      " doesn\n",
      " mean\n",
      " want\n",
      " open\n",
      " read\n",
      " make\n",
      " need\n",
      " ago\n",
      " thank\n",
      " \n",
      " \n",
      "Cluster 5: \n",
      " like\n",
      " looks\n",
      " sounds\n",
      " looks like\n",
      " sounds like\n",
      " things\n",
      " look\n",
      " feel\n",
      " way\n",
      " things like\n",
      " don\n",
      " just\n",
      " feel like\n",
      " lot\n",
      " make\n",
      " thing\n",
      " new\n",
      " just like\n",
      " doesn\n",
      " look like\n",
      " \n",
      " \n",
      "Cluster 6: \n",
      " windows\n",
      " using\n",
      " app\n",
      " linux\n",
      " web\n",
      " new\n",
      " software\n",
      " support\n",
      " apps\n",
      " desktop\n",
      " user\n",
      " os\n",
      " need\n",
      " run\n",
      " apple\n",
      " used\n",
      " users\n",
      " data\n",
      " make\n",
      " different\n",
      " \n",
      " \n",
      "Cluster 7: \n",
      " really\n",
      " just\n",
      " like\n",
      " didn\n",
      " doesn\n",
      " great\n",
      " interesting\n",
      " need\n",
      " think\n",
      " isn\n",
      " want\n",
      " people\n",
      " don\n",
      " maybe\n",
      " things\n",
      " point\n",
      " cool\n",
      " say\n",
      " ve\n",
      " pretty\n",
      " \n",
      " \n",
      "Cluster 8: \n",
      " use\n",
      " just\n",
      " used\n",
      " like\n",
      " case\n",
      " don\n",
      " want\n",
      " data\n",
      " using\n",
      " people\n",
      " language\n",
      " better\n",
      " don use\n",
      " time\n",
      " isn\n",
      " use case\n",
      " just use\n",
      " instead\n",
      " does\n",
      " app\n",
      " \n",
      " \n",
      "Cluster 9: \n",
      " way\n",
      " better\n",
      " great\n",
      " things\n",
      " sure\n",
      " make\n",
      " lot\n",
      " new\n",
      " pretty\n",
      " probably\n",
      " interesting\n",
      " isn\n",
      " going\n",
      " ll\n",
      " actually\n",
      " thing\n",
      " doing\n",
      " years\n",
      " ve\n",
      " day\n",
      " \n",
      " \n",
      "Cluster 10: \n",
      " wikipedia\n",
      " org\n",
      " wiki\n",
      " org wiki\n",
      " wikipedia org\n",
      " https en\n",
      " en wikipedia\n",
      " en\n",
      " https\n",
      " www\n",
      " like\n",
      " https www\n",
      " people\n",
      " http\n",
      " just\n",
      " article\n",
      " time\n",
      " way\n",
      " think\n",
      " used\n",
      " \n",
      " \n",
      "Cluster 11: \n",
      " article\n",
      " point\n",
      " problem\n",
      " isn\n",
      " way\n",
      " doesn\n",
      " thing\n",
      " used\n",
      " read\n",
      " data\n",
      " right\n",
      " actually\n",
      " different\n",
      " case\n",
      " didn\n",
      " say\n",
      " true\n",
      " comment\n",
      " said\n",
      " government\n",
      " \n",
      " \n",
      "Cluster 12: \n",
      " just\n",
      " like\n",
      " need\n",
      " doesn\n",
      " thing\n",
      " way\n",
      " just like\n",
      " don\n",
      " want\n",
      " different\n",
      " actually\n",
      " new\n",
      " point\n",
      " sure\n",
      " say\n",
      " probably\n",
      " isn\n",
      " did\n",
      " things\n",
      " know\n",
      " \n",
      " \n",
      "Cluster 13: \n",
      " right\n",
      " maybe\n",
      " say\n",
      " things\n",
      " make\n",
      " wrong\n",
      " just\n",
      " going\n",
      " long\n",
      " way\n",
      " like\n",
      " want\n",
      " left\n",
      " doing\n",
      " government\n",
      " problem\n",
      " world\n",
      " business\n",
      " bad\n",
      " think\n",
      " \n",
      " \n",
      "Cluster 14: \n",
      " good\n",
      " point\n",
      " idea\n",
      " pretty\n",
      " thing\n",
      " bad\n",
      " just\n",
      " like\n",
      " people\n",
      " better\n",
      " luck\n",
      " good idea\n",
      " good point\n",
      " think\n",
      " good luck\n",
      " pretty good\n",
      " job\n",
      " things\n",
      " make\n",
      " way\n",
      " \n",
      " \n",
      "Cluster 15: \n",
      " people\n",
      " like\n",
      " just\n",
      " lot\n",
      " want\n",
      " don\n",
      " think\n",
      " make\n",
      " know\n",
      " things\n",
      " problem\n",
      " say\n",
      " world\n",
      " actually\n",
      " money\n",
      " people don\n",
      " different\n",
      " better\n",
      " thing\n",
      " person\n",
      " \n",
      " \n",
      "Cluster 16: \n",
      " https\n",
      " com\n",
      " https www\n",
      " www\n",
      " github\n",
      " github com\n",
      " https github\n",
      " news\n",
      " ycombinator\n",
      " ycombinator com\n",
      " news ycombinator\n",
      " https news\n",
      " youtube com\n",
      " www youtube\n",
      " youtube\n",
      " id\n",
      " com watch\n",
      " item id\n",
      " com item\n",
      " item\n",
      " \n",
      " \n",
      "Cluster 17: \n",
      " time\n",
      " long\n",
      " point\n",
      " just\n",
      " long time\n",
      " lot\n",
      " like\n",
      " going\n",
      " ll\n",
      " new\n",
      " using\n",
      " things\n",
      " years\n",
      " day\n",
      " people\n",
      " think\n",
      " spend\n",
      " real\n",
      " working\n",
      " hours\n",
      " \n",
      " \n",
      "Cluster 18: \n",
      " think\n",
      " don think\n",
      " don\n",
      " people\n",
      " just\n",
      " like\n",
      " point\n",
      " way\n",
      " did\n",
      " better\n",
      " lot\n",
      " makes\n",
      " going\n",
      " different\n",
      " actually\n",
      " probably\n",
      " sure\n",
      " agree\n",
      " does\n",
      " things\n",
      " \n",
      " \n",
      "Cluster 19: \n",
      " don\n",
      " know\n",
      " don know\n",
      " want\n",
      " need\n",
      " don think\n",
      " like\n",
      " just\n",
      " understand\n",
      " people\n",
      " don want\n",
      " make\n",
      " way\n",
      " problem\n",
      " don need\n",
      " things\n",
      " care\n",
      " didn\n",
      " say\n",
      " doesn\n",
      " \n",
      " \n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"Top terms per cluster by Tfidf ranking:\")\n",
    "for i in range(number_of_clusters):\n",
    "    print(\"Cluster %d:\" % i, end='')\n",
    "    print(\" \")\n",
    "    for ind in order_centroids[i, :20]:\n",
    "        print(' %s' % terms[ind], end='')  # print phrases in clusters\n",
    "        print(\"\")\n",
    "    print(\" \")\n",
    "    print(\" \")\n",
    "print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Clarification of Clusters\n",
    "\n",
    "All the documents that typically have the words in a cluster are likely to be clustered together. But this doesn't detail anything about distances between words in documents. So we still don't know the meaning of these words or in what context they're used in sentences."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "What about other types clustering models?\n",
    "\n",
    "[sklearn documentation on agglomerative clustering](http://scikit-learn.org/stable/modules/generated/sklearn.cluster.AgglomerativeClustering.html)\n",
    "\n",
    "Hierarchical clustering (broader strategy of agglomerative clustering) is 0*n^2*log(n) which makes this operation too slow for large dataset such as mine..\n",
    "\n",
    "My HN comments are often very dissimilar so I don't think cosinse similarity would be very helpful.\n",
    "\n",
    "DBScan and Mini Batch KMeans both work well with a large number of samples. I could try those and compare results of clusters.\n",
    "\n",
    "[sklearn documentation on DBScan](http://scikit-learn.org/stable/modules/generated/sklearn.cluster.DBSCAN.html)\n",
    "\n",
    "[sklearn documentation on Mini Batch K Means](http://scikit-learn.org/stable/modules/generated/sklearn.cluster.MiniBatchKMeans.html)\n",
    "\n",
    "Lessons learned: DBScan took an incredibly long amount of time to run. It may not be best for my large number of samples - especially because I'd want to utilize grid search on DBScan to optimally find clusters and can't without GPUs or a higher performance computer."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Mini Batch K Means is incredibly fast! "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "mini_batch_kmeans = MiniBatchKMeans(init='k-means++', init_size=1000,\n",
    "                                   batch_size=200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MiniBatchKMeans(batch_size=200, compute_labels=True, init='k-means++',\n",
       "        init_size=1000, max_iter=100, max_no_improvement=10, n_clusters=8,\n",
       "        n_init=3, random_state=None, reassignment_ratio=0.01, tol=0.0,\n",
       "        verbose=0)"
      ]
     },
     "execution_count": 105,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mini_batch_kmeans.fit(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "original_space_centroids = svd.inverse_transform(km.cluster_centers_)\n",
    "order_centroids = original_space_centroids.argsort()[:, ::-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "terms = vectorizer.get_feature_names()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top terms per cluster by Tfidf ranking:\n",
      "Cluster 0: \n",
      " ve\n",
      " seen\n",
      " ve seen\n",
      " got\n",
      " used\n",
      " heard\n",
      " ve heard\n",
      " years\n",
      " just\n",
      " read\n",
      " did\n",
      " actually\n",
      " things\n",
      " ve got\n",
      " like\n",
      " people\n",
      " lot\n",
      " don\n",
      " using\n",
      " probably\n",
      " \n",
      " \n",
      "Cluster 1: \n",
      " work\n",
      " job\n",
      " working\n",
      " day\n",
      " company\n",
      " doesn\n",
      " time\n",
      " people\n",
      " just\n",
      " hours\n",
      " like\n",
      " doing\n",
      " need\n",
      " hard\n",
      " way\n",
      " lot\n",
      " home\n",
      " does\n",
      " think\n",
      " pay\n",
      " \n",
      " \n",
      "Cluster 2: \n",
      " http\n",
      " http www\n",
      " com\n",
      " www\n",
      " org\n",
      " news\n",
      " html\n",
      " article\n",
      " like\n",
      " uk\n",
      " net\n",
      " blog\n",
      " data\n",
      " app\n",
      " just\n",
      " way\n",
      " did\n",
      " https www\n",
      " doesn\n",
      " people\n",
      " \n",
      " \n",
      "Cluster 3: \n",
      " money\n",
      " uber\n",
      " company\n",
      " pay\n",
      " market\n",
      " companies\n",
      " business\n",
      " tax\n",
      " new\n",
      " cost\n",
      " year\n",
      " make\n",
      " price\n",
      " government\n",
      " free\n",
      " service\n",
      " going\n",
      " car\n",
      " high\n",
      " product\n",
      " \n",
      " \n",
      "Cluster 4: \n",
      " does\n",
      " google\n",
      " code\n",
      " thanks\n",
      " yes\n",
      " data\n",
      " years\n",
      " source\n",
      " did\n",
      " article\n",
      " language\n",
      " doesn\n",
      " mean\n",
      " want\n",
      " open\n",
      " read\n",
      " make\n",
      " need\n",
      " ago\n",
      " thank\n",
      " \n",
      " \n",
      "Cluster 5: \n",
      " like\n",
      " looks\n",
      " sounds\n",
      " looks like\n",
      " sounds like\n",
      " things\n",
      " look\n",
      " feel\n",
      " way\n",
      " things like\n",
      " don\n",
      " just\n",
      " feel like\n",
      " lot\n",
      " make\n",
      " thing\n",
      " new\n",
      " just like\n",
      " doesn\n",
      " look like\n",
      " \n",
      " \n",
      "Cluster 6: \n",
      " windows\n",
      " using\n",
      " app\n",
      " linux\n",
      " web\n",
      " new\n",
      " software\n",
      " support\n",
      " apps\n",
      " desktop\n",
      " user\n",
      " os\n",
      " need\n",
      " run\n",
      " apple\n",
      " used\n",
      " users\n",
      " data\n",
      " make\n",
      " different\n",
      " \n",
      " \n",
      "Cluster 7: \n",
      " really\n",
      " just\n",
      " like\n",
      " didn\n",
      " doesn\n",
      " great\n",
      " interesting\n",
      " need\n",
      " think\n",
      " isn\n",
      " want\n",
      " people\n",
      " don\n",
      " maybe\n",
      " things\n",
      " point\n",
      " cool\n",
      " say\n",
      " ve\n",
      " pretty\n",
      " \n",
      " \n",
      "Cluster 8: \n",
      " use\n",
      " just\n",
      " used\n",
      " like\n",
      " case\n",
      " don\n",
      " want\n",
      " data\n",
      " using\n",
      " people\n",
      " language\n",
      " better\n",
      " don use\n",
      " time\n",
      " isn\n",
      " use case\n",
      " just use\n",
      " instead\n",
      " does\n",
      " app\n",
      " \n",
      " \n",
      "Cluster 9: \n",
      " way\n",
      " better\n",
      " great\n",
      " things\n",
      " sure\n",
      " make\n",
      " lot\n",
      " new\n",
      " pretty\n",
      " probably\n",
      " interesting\n",
      " isn\n",
      " going\n",
      " ll\n",
      " actually\n",
      " thing\n",
      " doing\n",
      " years\n",
      " ve\n",
      " day\n",
      " \n",
      " \n",
      "Cluster 10: \n",
      " wikipedia\n",
      " org\n",
      " wiki\n",
      " org wiki\n",
      " wikipedia org\n",
      " https en\n",
      " en wikipedia\n",
      " en\n",
      " https\n",
      " www\n",
      " like\n",
      " https www\n",
      " people\n",
      " http\n",
      " just\n",
      " article\n",
      " time\n",
      " way\n",
      " think\n",
      " used\n",
      " \n",
      " \n",
      "Cluster 11: \n",
      " article\n",
      " point\n",
      " problem\n",
      " isn\n",
      " way\n",
      " doesn\n",
      " thing\n",
      " used\n",
      " read\n",
      " data\n",
      " right\n",
      " actually\n",
      " different\n",
      " case\n",
      " didn\n",
      " say\n",
      " true\n",
      " comment\n",
      " said\n",
      " government\n",
      " \n",
      " \n",
      "Cluster 12: \n",
      " just\n",
      " like\n",
      " need\n",
      " doesn\n",
      " thing\n",
      " way\n",
      " just like\n",
      " don\n",
      " want\n",
      " different\n",
      " actually\n",
      " new\n",
      " point\n",
      " sure\n",
      " say\n",
      " probably\n",
      " isn\n",
      " did\n",
      " things\n",
      " know\n",
      " \n",
      " \n",
      "Cluster 13: \n",
      " right\n",
      " maybe\n",
      " say\n",
      " things\n",
      " make\n",
      " wrong\n",
      " just\n",
      " going\n",
      " long\n",
      " way\n",
      " like\n",
      " want\n",
      " left\n",
      " doing\n",
      " government\n",
      " problem\n",
      " world\n",
      " business\n",
      " bad\n",
      " think\n",
      " \n",
      " \n",
      "Cluster 14: \n",
      " good\n",
      " point\n",
      " idea\n",
      " pretty\n",
      " thing\n",
      " bad\n",
      " just\n",
      " like\n",
      " people\n",
      " better\n",
      " luck\n",
      " good idea\n",
      " good point\n",
      " think\n",
      " good luck\n",
      " pretty good\n",
      " job\n",
      " things\n",
      " make\n",
      " way\n",
      " \n",
      " \n",
      "Cluster 15: \n",
      " people\n",
      " like\n",
      " just\n",
      " lot\n",
      " want\n",
      " don\n",
      " think\n",
      " make\n",
      " know\n",
      " things\n",
      " problem\n",
      " say\n",
      " world\n",
      " actually\n",
      " money\n",
      " people don\n",
      " different\n",
      " better\n",
      " thing\n",
      " person\n",
      " \n",
      " \n",
      "Cluster 16: \n",
      " https\n",
      " com\n",
      " https www\n",
      " www\n",
      " github\n",
      " github com\n",
      " https github\n",
      " news\n",
      " ycombinator\n",
      " ycombinator com\n",
      " news ycombinator\n",
      " https news\n",
      " youtube com\n",
      " www youtube\n",
      " youtube\n",
      " id\n",
      " com watch\n",
      " item id\n",
      " com item\n",
      " item\n",
      " \n",
      " \n",
      "Cluster 17: \n",
      " time\n",
      " long\n",
      " point\n",
      " just\n",
      " long time\n",
      " lot\n",
      " like\n",
      " going\n",
      " ll\n",
      " new\n",
      " using\n",
      " things\n",
      " years\n",
      " day\n",
      " people\n",
      " think\n",
      " spend\n",
      " real\n",
      " working\n",
      " hours\n",
      " \n",
      " \n",
      "Cluster 18: \n",
      " think\n",
      " don think\n",
      " don\n",
      " people\n",
      " just\n",
      " like\n",
      " point\n",
      " way\n",
      " did\n",
      " better\n",
      " lot\n",
      " makes\n",
      " going\n",
      " different\n",
      " actually\n",
      " probably\n",
      " sure\n",
      " agree\n",
      " does\n",
      " things\n",
      " \n",
      " \n",
      "Cluster 19: \n",
      " don\n",
      " know\n",
      " don know\n",
      " want\n",
      " need\n",
      " don think\n",
      " like\n",
      " just\n",
      " understand\n",
      " people\n",
      " don want\n",
      " make\n",
      " way\n",
      " problem\n",
      " don need\n",
      " things\n",
      " care\n",
      " didn\n",
      " say\n",
      " doesn\n",
      " \n",
      " \n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"Top terms per cluster by Tfidf ranking:\")\n",
    "for i in range(number_of_clusters):\n",
    "    print(\"Cluster %d:\" % i, end='')\n",
    "    print(\" \")\n",
    "    for ind in order_centroids[i, :20]:\n",
    "        print(' %s' % terms[ind], end='')  # print phrases in clusters\n",
    "        print(\"\")\n",
    "    print(\" \")\n",
    "    print(\" \")\n",
    "print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ultimately, these clusters included words that often didn't appear as top words in other clusters. Therefore, they were fairly distinguished from one another. The only issue is that clustering used a bag of words approach. So words like \"use\", \"like\" and \"master\" are vague because I don't know in what context they appeared. There are certainly better ways we can get insights from these Hacker News comments."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Topic Extraction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### LDA"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Latent Dirichlet Allocation (LDA) is a way of discovering topics that documents contain. You pass, as a parameter, a value for the number of topics you want to create. \n",
    "\n",
    "LDA is a learning model.\n",
    "\n",
    "With each topic, you can see words and probabilities associated with seeing each of those words across all the Hacker News comments. You don't get an exact topic name; but based on words like \"computer, python and java\", you can infer the topic is about computer programming.\n",
    "\n",
    "Great [Quora post on LDA](https://www.quora.com/What-is-a-good-explanation-of-Latent-Dirichlet-Allocation)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[SKlearn documentation on LDA](http://scikit-learn.org/stable/modules/generated/sklearn.decomposition.LatentDirichletAllocation.html)\n",
    "\n",
    "I tried using LDA through SKlearn but the process took a very long time and never finished. Instead, I've heard Gensim is much better performance-wise.\n",
    "\n",
    "[Gensim documentation on LDA](https://radimrehurek.com/gensim/models/ldamodel.html)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "With LDA, we map text documents from a word space to a topic space - that gives us the topic distribution of different documents in our corpus; it also allows us to make conceptual comparisions of documents in this reduced topic space."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Topic Modeling with Gensim"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We transpose our Count Vectorizer matrix so the terms are rows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "counts = count_vect_matrix.transpose()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(83175, 246175)"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "counts.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We need to convert our sparse scipy matrix to a gensim-friendly objected called a Corpus."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "corpus = matutils.Sparse2Corpus(counts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "gensim.matutils.Sparse2Corpus"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "type(corpus)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The term \"flying car\" is at feature index 25052"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "25052"
      ]
     },
     "execution_count": 113,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "count_vectorizer.vocabulary_['flying car']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We need to save a mapping (dict) of row id to word (token) for later use by gensim. We call the variable id2word but that can include bigrams - as seen above in \"flying car\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "id2word = dict((v, k) for k, v in count_vectorizer.vocabulary_.items())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "83175"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(id2word)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Create our LDA model. It requires our corpus of word counts, mapping of row ids to words, and the number of topics. \n",
    "\n",
    "We're doing unsupervised learning so there's no perfect answer on the number of topics we need. I'll start with 15."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2017-06-05 20:56:09,339 : INFO : using symmetric alpha at 0.06666666666666667\n",
      "2017-06-05 20:56:09,340 : INFO : using symmetric eta at 1.20228434025e-05\n",
      "2017-06-05 20:56:09,356 : INFO : using serial LDA version on this node\n",
      "2017-06-05 20:56:15,417 : INFO : running online LDA training, 15 topics, 1 passes over the supplied corpus of 246175 documents, updating model once every 2000 documents, evaluating perplexity every 20000 documents, iterating 50x with a convergence threshold of 0.001000\n",
      "2017-06-05 20:56:15,451 : INFO : PROGRESS: pass 0, at document #2000/246175\n",
      "2017-06-05 20:56:18,249 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:18,444 : INFO : topic #10 (0.067): 0.005*\"really\" + 0.005*\"way\" + 0.005*\"just\" + 0.004*\"https\" + 0.004*\"use\" + 0.004*\"com\" + 0.003*\"want\" + 0.003*\"like\" + 0.003*\"time\" + 0.003*\"right\"\n",
      "2017-06-05 20:56:18,446 : INFO : topic #7 (0.067): 0.006*\"just\" + 0.005*\"like\" + 0.005*\"https\" + 0.004*\"don\" + 0.004*\"people\" + 0.004*\"use\" + 0.004*\"way\" + 0.003*\"com\" + 0.003*\"better\" + 0.003*\"work\"\n",
      "2017-06-05 20:56:18,448 : INFO : topic #1 (0.067): 0.009*\"people\" + 0.007*\"like\" + 0.006*\"just\" + 0.005*\"don\" + 0.005*\"make\" + 0.004*\"really\" + 0.004*\"com\" + 0.003*\"know\" + 0.003*\"way\" + 0.003*\"tax\"\n",
      "2017-06-05 20:56:18,450 : INFO : topic #9 (0.067): 0.007*\"tax\" + 0.006*\"just\" + 0.006*\"like\" + 0.005*\"good\" + 0.005*\"time\" + 0.005*\"think\" + 0.005*\"don\" + 0.004*\"taxes\" + 0.004*\"people\" + 0.003*\"work\"\n",
      "2017-06-05 20:56:18,452 : INFO : topic #8 (0.067): 0.008*\"like\" + 0.006*\"https\" + 0.006*\"just\" + 0.006*\"com\" + 0.005*\"google\" + 0.005*\"think\" + 0.004*\"http\" + 0.004*\"people\" + 0.004*\"don\" + 0.004*\"www\"\n",
      "2017-06-05 20:56:18,459 : INFO : topic diff=15.483290, rho=1.000000\n",
      "2017-06-05 20:56:18,506 : INFO : PROGRESS: pass 0, at document #4000/246175\n",
      "2017-06-05 20:56:20,801 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:21,032 : INFO : topic #3 (0.067): 0.006*\"people\" + 0.006*\"money\" + 0.005*\"don\" + 0.005*\"just\" + 0.005*\"doesn\" + 0.005*\"work\" + 0.004*\"use\" + 0.004*\"does\" + 0.004*\"ve\" + 0.004*\"think\"\n",
      "2017-06-05 20:56:21,033 : INFO : topic #5 (0.067): 0.008*\"just\" + 0.008*\"https\" + 0.008*\"like\" + 0.006*\"com\" + 0.005*\"browser\" + 0.005*\"really\" + 0.005*\"windows\" + 0.004*\"people\" + 0.004*\"ve\" + 0.004*\"time\"\n",
      "2017-06-05 20:56:21,036 : INFO : topic #4 (0.067): 0.008*\"people\" + 0.005*\"like\" + 0.005*\"don\" + 0.004*\"think\" + 0.004*\"time\" + 0.004*\"way\" + 0.004*\"just\" + 0.004*\"use\" + 0.003*\"power\" + 0.003*\"point\"\n",
      "2017-06-05 20:56:21,038 : INFO : topic #8 (0.067): 0.013*\"union\" + 0.008*\"like\" + 0.006*\"unions\" + 0.005*\"don\" + 0.005*\"people\" + 0.005*\"think\" + 0.005*\"google\" + 0.005*\"just\" + 0.005*\"workers\" + 0.005*\"https\"\n",
      "2017-06-05 20:56:21,040 : INFO : topic #6 (0.067): 0.013*\"people\" + 0.011*\"just\" + 0.010*\"think\" + 0.008*\"don\" + 0.007*\"like\" + 0.006*\"blockchain\" + 0.006*\"way\" + 0.005*\"really\" + 0.005*\"make\" + 0.004*\"using\"\n",
      "2017-06-05 20:56:21,047 : INFO : topic diff=1.096247, rho=0.707107\n",
      "2017-06-05 20:56:21,084 : INFO : PROGRESS: pass 0, at document #6000/246175\n",
      "2017-06-05 20:56:23,219 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:23,567 : INFO : topic #14 (0.067): 0.009*\"like\" + 0.006*\"use\" + 0.005*\"don\" + 0.005*\"just\" + 0.005*\"used\" + 0.005*\"java\" + 0.004*\"archive\" + 0.004*\"change\" + 0.003*\"really\" + 0.003*\"ve\"\n",
      "2017-06-05 20:56:23,569 : INFO : topic #13 (0.067): 0.009*\"don\" + 0.007*\"just\" + 0.006*\"people\" + 0.006*\"problem\" + 0.005*\"survivorship\" + 0.005*\"things\" + 0.005*\"bias\" + 0.005*\"survivorship bias\" + 0.005*\"like\" + 0.005*\"think\"\n",
      "2017-06-05 20:56:23,571 : INFO : topic #2 (0.067): 0.008*\"just\" + 0.007*\"don\" + 0.006*\"people\" + 0.005*\"money\" + 0.005*\"work\" + 0.005*\"make\" + 0.005*\"use\" + 0.005*\"doesn\" + 0.005*\"time\" + 0.004*\"good\"\n",
      "2017-06-05 20:56:23,574 : INFO : topic #1 (0.067): 0.012*\"people\" + 0.007*\"don\" + 0.006*\"just\" + 0.005*\"make\" + 0.005*\"like\" + 0.004*\"think\" + 0.004*\"doesn\" + 0.004*\"stem\" + 0.004*\"bitcoin\" + 0.004*\"really\"\n",
      "2017-06-05 20:56:23,576 : INFO : topic #9 (0.067): 0.007*\"good\" + 0.007*\"gender\" + 0.006*\"just\" + 0.006*\"don\" + 0.006*\"id\" + 0.005*\"item\" + 0.005*\"ycombinator com\" + 0.005*\"people\" + 0.005*\"ycombinator\" + 0.005*\"news ycombinator\"\n",
      "2017-06-05 20:56:23,584 : INFO : topic diff=0.784957, rho=0.577350\n",
      "2017-06-05 20:56:23,609 : INFO : PROGRESS: pass 0, at document #8000/246175\n",
      "2017-06-05 20:56:25,686 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:26,131 : INFO : topic #6 (0.067): 0.013*\"just\" + 0.012*\"people\" + 0.010*\"think\" + 0.009*\"don\" + 0.007*\"like\" + 0.007*\"way\" + 0.006*\"really\" + 0.005*\"using\" + 0.005*\"work\" + 0.004*\"make\"\n",
      "2017-06-05 20:56:26,133 : INFO : topic #11 (0.067): 0.016*\"svg\" + 0.007*\"http\" + 0.007*\"com\" + 0.006*\"people\" + 0.006*\"like\" + 0.006*\"www\" + 0.005*\"clothes\" + 0.005*\"think\" + 0.005*\"just\" + 0.005*\"https\"\n",
      "2017-06-05 20:56:26,135 : INFO : topic #10 (0.067): 0.007*\"use\" + 0.007*\"really\" + 0.005*\"just\" + 0.005*\"pain\" + 0.005*\"code\" + 0.005*\"book\" + 0.005*\"way\" + 0.004*\"want\" + 0.004*\"different\" + 0.004*\"like\"\n",
      "2017-06-05 20:56:26,137 : INFO : topic #7 (0.067): 0.013*\"women\" + 0.005*\"like\" + 0.005*\"just\" + 0.004*\"say\" + 0.004*\"don\" + 0.004*\"society\" + 0.004*\"think\" + 0.003*\"years\" + 0.003*\"people\" + 0.003*\"https\"\n",
      "2017-06-05 20:56:26,140 : INFO : topic #12 (0.067): 0.014*\"com\" + 0.013*\"like\" + 0.011*\"https\" + 0.009*\"www\" + 0.006*\"https www\" + 0.006*\"time\" + 0.005*\"google\" + 0.005*\"ago\" + 0.005*\"just\" + 0.004*\"think\"\n",
      "2017-06-05 20:56:26,147 : INFO : topic diff=0.716134, rho=0.500000\n",
      "2017-06-05 20:56:26,176 : INFO : PROGRESS: pass 0, at document #10000/246175\n",
      "2017-06-05 20:56:28,265 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:28,838 : INFO : topic #12 (0.067): 0.015*\"com\" + 0.013*\"like\" + 0.012*\"https\" + 0.010*\"www\" + 0.007*\"vim\" + 0.007*\"https www\" + 0.006*\"google\" + 0.006*\"time\" + 0.005*\"ago\" + 0.005*\"years\"\n",
      "2017-06-05 20:56:28,841 : INFO : topic #1 (0.067): 0.014*\"people\" + 0.008*\"don\" + 0.007*\"card\" + 0.007*\"like\" + 0.006*\"just\" + 0.005*\"make\" + 0.005*\"cards\" + 0.004*\"really\" + 0.004*\"doesn\" + 0.004*\"know\"\n",
      "2017-06-05 20:56:28,843 : INFO : topic #3 (0.067): 0.017*\"docker\" + 0.006*\"work\" + 0.006*\"use\" + 0.006*\"just\" + 0.006*\"don\" + 0.005*\"accounts\" + 0.005*\"different\" + 0.005*\"people\" + 0.004*\"does\" + 0.004*\"need\"\n",
      "2017-06-05 20:56:28,846 : INFO : topic #4 (0.067): 0.007*\"people\" + 0.006*\"facebook\" + 0.006*\"social\" + 0.005*\"like\" + 0.005*\"don\" + 0.005*\"movie\" + 0.004*\"think\" + 0.004*\"just\" + 0.004*\"use\" + 0.004*\"way\"\n",
      "2017-06-05 20:56:28,848 : INFO : topic #6 (0.067): 0.013*\"just\" + 0.013*\"people\" + 0.010*\"think\" + 0.010*\"don\" + 0.007*\"like\" + 0.007*\"way\" + 0.006*\"really\" + 0.005*\"work\" + 0.005*\"python\" + 0.005*\"using\"\n",
      "2017-06-05 20:56:28,855 : INFO : topic diff=0.728102, rho=0.447214\n",
      "2017-06-05 20:56:28,882 : INFO : PROGRESS: pass 0, at document #12000/246175\n",
      "2017-06-05 20:56:30,914 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:31,600 : INFO : topic #8 (0.067): 0.009*\"like\" + 0.006*\"google\" + 0.005*\"people\" + 0.005*\"don\" + 0.005*\"just\" + 0.004*\"think\" + 0.004*\"good\" + 0.004*\"bad\" + 0.004*\"really\" + 0.003*\"https\"\n",
      "2017-06-05 20:56:31,602 : INFO : topic #7 (0.067): 0.005*\"like\" + 0.005*\"women\" + 0.004*\"united\" + 0.004*\"just\" + 0.004*\"say\" + 0.004*\"don\" + 0.004*\"people\" + 0.004*\"use\" + 0.004*\"comment\" + 0.004*\"think\"\n",
      "2017-06-05 20:56:31,605 : INFO : topic #11 (0.067): 0.008*\"dune\" + 0.007*\"http\" + 0.007*\"like\" + 0.006*\"people\" + 0.006*\"www\" + 0.006*\"http www\" + 0.005*\"svg\" + 0.005*\"com\" + 0.005*\"think\" + 0.005*\"just\"\n",
      "2017-06-05 20:56:31,607 : INFO : topic #0 (0.067): 0.009*\"people\" + 0.007*\"drugs\" + 0.007*\"com\" + 0.006*\"https\" + 0.005*\"airline\" + 0.005*\"don\" + 0.005*\"just\" + 0.004*\"point\" + 0.004*\"business\" + 0.004*\"blog\"\n",
      "2017-06-05 20:56:31,609 : INFO : topic #6 (0.067): 0.013*\"just\" + 0.013*\"people\" + 0.011*\"think\" + 0.010*\"don\" + 0.008*\"like\" + 0.007*\"way\" + 0.007*\"really\" + 0.006*\"work\" + 0.005*\"using\" + 0.005*\"need\"\n",
      "2017-06-05 20:56:31,616 : INFO : topic diff=0.586344, rho=0.408248\n",
      "2017-06-05 20:56:31,642 : INFO : PROGRESS: pass 0, at document #14000/246175\n",
      "2017-06-05 20:56:33,672 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:34,446 : INFO : topic #14 (0.067): 0.011*\"like\" + 0.008*\"use\" + 0.008*\"java\" + 0.005*\"just\" + 0.004*\"don\" + 0.004*\"code\" + 0.004*\"language\" + 0.004*\"change\" + 0.004*\"ve\" + 0.004*\"data\"\n",
      "2017-06-05 20:56:34,448 : INFO : topic #10 (0.067): 0.009*\"use\" + 0.008*\"code\" + 0.007*\"really\" + 0.006*\"just\" + 0.005*\"js\" + 0.005*\"want\" + 0.004*\"javascript\" + 0.004*\"like\" + 0.004*\"way\" + 0.004*\"need\"\n",
      "2017-06-05 20:56:34,450 : INFO : topic #8 (0.067): 0.009*\"like\" + 0.007*\"google\" + 0.005*\"people\" + 0.005*\"don\" + 0.005*\"think\" + 0.004*\"just\" + 0.004*\"bad\" + 0.004*\"good\" + 0.004*\"airlines\" + 0.004*\"really\"\n",
      "2017-06-05 20:56:34,452 : INFO : topic #12 (0.067): 0.017*\"com\" + 0.014*\"like\" + 0.012*\"https\" + 0.011*\"www\" + 0.011*\"vim\" + 0.008*\"https www\" + 0.006*\"time\" + 0.006*\"years\" + 0.006*\"google\" + 0.005*\"lidar\"\n",
      "2017-06-05 20:56:34,454 : INFO : topic #4 (0.067): 0.007*\"people\" + 0.006*\"laptop\" + 0.005*\"like\" + 0.005*\"don\" + 0.004*\"facebook\" + 0.004*\"source\" + 0.004*\"use\" + 0.004*\"think\" + 0.004*\"social\" + 0.004*\"movie\"\n",
      "2017-06-05 20:56:34,461 : INFO : topic diff=0.581639, rho=0.377964\n",
      "2017-06-05 20:56:34,487 : INFO : PROGRESS: pass 0, at document #16000/246175\n",
      "2017-06-05 20:56:36,409 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:37,213 : INFO : topic #12 (0.067): 0.019*\"com\" + 0.014*\"like\" + 0.014*\"https\" + 0.013*\"www\" + 0.010*\"https www\" + 0.008*\"vim\" + 0.007*\"years\" + 0.007*\"time\" + 0.006*\"http\" + 0.006*\"google\"\n",
      "2017-06-05 20:56:37,215 : INFO : topic #3 (0.067): 0.012*\"docker\" + 0.006*\"use\" + 0.006*\"work\" + 0.006*\"don\" + 0.006*\"just\" + 0.005*\"does\" + 0.005*\"need\" + 0.005*\"linux\" + 0.004*\"google\" + 0.004*\"want\"\n",
      "2017-06-05 20:56:37,218 : INFO : topic #13 (0.067): 0.008*\"don\" + 0.007*\"just\" + 0.007*\"problem\" + 0.006*\"like\" + 0.006*\"people\" + 0.005*\"things\" + 0.005*\"google\" + 0.005*\"software\" + 0.004*\"code\" + 0.004*\"bbs\"\n",
      "2017-06-05 20:56:37,219 : INFO : topic #5 (0.067): 0.017*\"https\" + 0.010*\"org\" + 0.009*\"like\" + 0.009*\"just\" + 0.008*\"com\" + 0.008*\"github\" + 0.007*\"wikipedia\" + 0.006*\"use\" + 0.006*\"web\" + 0.006*\"windows\"\n",
      "2017-06-05 20:56:37,221 : INFO : topic #1 (0.067): 0.015*\"people\" + 0.008*\"don\" + 0.006*\"like\" + 0.005*\"make\" + 0.005*\"just\" + 0.004*\"know\" + 0.004*\"really\" + 0.004*\"think\" + 0.004*\"doesn\" + 0.004*\"understand\"\n",
      "2017-06-05 20:56:37,229 : INFO : topic diff=0.537071, rho=0.353553\n",
      "2017-06-05 20:56:37,257 : INFO : PROGRESS: pass 0, at document #18000/246175\n",
      "2017-06-05 20:56:39,265 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:40,076 : INFO : topic #14 (0.067): 0.012*\"like\" + 0.008*\"use\" + 0.006*\"java\" + 0.006*\"data\" + 0.005*\"just\" + 0.004*\"code\" + 0.004*\"don\" + 0.004*\"language\" + 0.004*\"used\" + 0.004*\"change\"\n",
      "2017-06-05 20:56:40,079 : INFO : topic #1 (0.067): 0.018*\"people\" + 0.009*\"don\" + 0.006*\"like\" + 0.006*\"make\" + 0.005*\"just\" + 0.005*\"know\" + 0.005*\"think\" + 0.005*\"really\" + 0.004*\"understand\" + 0.004*\"doesn\"\n",
      "2017-06-05 20:56:40,081 : INFO : topic #11 (0.067): 0.008*\"http\" + 0.006*\"like\" + 0.006*\"www\" + 0.006*\"http www\" + 0.005*\"people\" + 0.005*\"police\" + 0.005*\"article\" + 0.005*\"think\" + 0.004*\"com\" + 0.004*\"just\"\n",
      "2017-06-05 20:56:40,083 : INFO : topic #5 (0.067): 0.018*\"https\" + 0.010*\"org\" + 0.009*\"just\" + 0.009*\"com\" + 0.009*\"like\" + 0.008*\"github\" + 0.008*\"windows\" + 0.006*\"wikipedia\" + 0.006*\"use\" + 0.006*\"github com\"\n",
      "2017-06-05 20:56:40,085 : INFO : topic #6 (0.067): 0.014*\"just\" + 0.013*\"people\" + 0.012*\"think\" + 0.012*\"don\" + 0.008*\"like\" + 0.007*\"work\" + 0.007*\"really\" + 0.007*\"need\" + 0.007*\"way\" + 0.006*\"language\"\n",
      "2017-06-05 20:56:40,091 : INFO : topic diff=0.560788, rho=0.333333\n",
      "2017-06-05 20:56:44,726 : INFO : -10.289 per-word bound, 1250.9 perplexity estimate based on a held-out corpus of 2000 documents with 68894 words\n",
      "2017-06-05 20:56:44,727 : INFO : PROGRESS: pass 0, at document #20000/246175\n",
      "2017-06-05 20:56:46,632 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:47,451 : INFO : topic #7 (0.067): 0.015*\"war\" + 0.005*\"labour\" + 0.005*\"like\" + 0.004*\"years\" + 0.004*\"france\" + 0.004*\"just\" + 0.004*\"say\" + 0.004*\"united\" + 0.004*\"did\" + 0.003*\"people\"\n",
      "2017-06-05 20:56:47,453 : INFO : topic #10 (0.067): 0.009*\"use\" + 0.008*\"code\" + 0.006*\"really\" + 0.006*\"just\" + 0.005*\"js\" + 0.004*\"want\" + 0.004*\"ipv\" + 0.004*\"like\" + 0.004*\"memory\" + 0.004*\"ve\"\n",
      "2017-06-05 20:56:47,455 : INFO : topic #6 (0.067): 0.014*\"just\" + 0.014*\"people\" + 0.012*\"think\" + 0.012*\"don\" + 0.008*\"like\" + 0.008*\"work\" + 0.007*\"really\" + 0.007*\"need\" + 0.007*\"way\" + 0.006*\"language\"\n",
      "2017-06-05 20:56:47,456 : INFO : topic #12 (0.067): 0.022*\"com\" + 0.015*\"https\" + 0.015*\"www\" + 0.015*\"like\" + 0.011*\"https www\" + 0.007*\"time\" + 0.007*\"years\" + 0.007*\"http\" + 0.007*\"ago\" + 0.005*\"day\"\n",
      "2017-06-05 20:56:47,458 : INFO : topic #1 (0.067): 0.019*\"people\" + 0.008*\"don\" + 0.006*\"like\" + 0.006*\"make\" + 0.005*\"know\" + 0.005*\"just\" + 0.005*\"think\" + 0.005*\"really\" + 0.005*\"understand\" + 0.004*\"doesn\"\n",
      "2017-06-05 20:56:47,465 : INFO : topic diff=0.494955, rho=0.316228\n",
      "2017-06-05 20:56:47,490 : INFO : PROGRESS: pass 0, at document #22000/246175\n",
      "2017-06-05 20:56:49,243 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:50,084 : INFO : topic #6 (0.067): 0.015*\"just\" + 0.014*\"people\" + 0.013*\"think\" + 0.012*\"don\" + 0.008*\"like\" + 0.008*\"work\" + 0.007*\"really\" + 0.007*\"need\" + 0.007*\"way\" + 0.006*\"language\"\n",
      "2017-06-05 20:56:50,085 : INFO : topic #12 (0.067): 0.024*\"com\" + 0.017*\"https\" + 0.017*\"www\" + 0.015*\"like\" + 0.012*\"https www\" + 0.008*\"http\" + 0.008*\"years\" + 0.007*\"ago\" + 0.007*\"time\" + 0.005*\"day\"\n",
      "2017-06-05 20:56:50,087 : INFO : topic #10 (0.067): 0.009*\"use\" + 0.007*\"code\" + 0.007*\"js\" + 0.006*\"really\" + 0.006*\"just\" + 0.005*\"want\" + 0.004*\"like\" + 0.004*\"memory\" + 0.004*\"ve\" + 0.004*\"way\"\n",
      "2017-06-05 20:56:50,089 : INFO : topic #0 (0.067): 0.011*\"people\" + 0.008*\"companies\" + 0.005*\"business\" + 0.005*\"just\" + 0.005*\"indian\" + 0.005*\"active\" + 0.005*\"index\" + 0.004*\"point\" + 0.004*\"don\" + 0.004*\"stock\"\n",
      "2017-06-05 20:56:50,091 : INFO : topic #7 (0.067): 0.012*\"war\" + 0.005*\"labour\" + 0.004*\"like\" + 0.004*\"say\" + 0.004*\"just\" + 0.004*\"women\" + 0.004*\"years\" + 0.004*\"united\" + 0.004*\"did\" + 0.004*\"think\"\n",
      "2017-06-05 20:56:50,098 : INFO : topic diff=0.434326, rho=0.301511\n",
      "2017-06-05 20:56:50,123 : INFO : PROGRESS: pass 0, at document #24000/246175\n",
      "2017-06-05 20:56:51,884 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:52,703 : INFO : topic #14 (0.067): 0.013*\"like\" + 0.010*\"data\" + 0.008*\"use\" + 0.005*\"java\" + 0.005*\"just\" + 0.004*\"code\" + 0.004*\"change\" + 0.004*\"don\" + 0.004*\"things\" + 0.004*\"used\"\n",
      "2017-06-05 20:56:52,705 : INFO : topic #3 (0.067): 0.008*\"docker\" + 0.007*\"funds\" + 0.006*\"work\" + 0.006*\"does\" + 0.006*\"don\" + 0.006*\"just\" + 0.005*\"use\" + 0.005*\"vanguard\" + 0.005*\"need\" + 0.005*\"different\"\n",
      "2017-06-05 20:56:52,707 : INFO : topic #13 (0.067): 0.008*\"just\" + 0.008*\"don\" + 0.008*\"things\" + 0.008*\"problem\" + 0.007*\"like\" + 0.006*\"people\" + 0.005*\"time\" + 0.005*\"thanks\" + 0.005*\"software\" + 0.004*\"work\"\n",
      "2017-06-05 20:56:52,710 : INFO : topic #8 (0.067): 0.008*\"like\" + 0.006*\"people\" + 0.005*\"think\" + 0.005*\"bad\" + 0.004*\"workers\" + 0.004*\"just\" + 0.004*\"don\" + 0.004*\"age\" + 0.004*\"power\" + 0.004*\"good\"\n",
      "2017-06-05 20:56:52,712 : INFO : topic #1 (0.067): 0.020*\"people\" + 0.008*\"don\" + 0.006*\"make\" + 0.006*\"like\" + 0.005*\"know\" + 0.005*\"just\" + 0.005*\"understand\" + 0.005*\"think\" + 0.005*\"doesn\" + 0.004*\"really\"\n",
      "2017-06-05 20:56:52,719 : INFO : topic diff=0.424270, rho=0.288675\n",
      "2017-06-05 20:56:52,744 : INFO : PROGRESS: pass 0, at document #26000/246175\n",
      "2017-06-05 20:56:54,552 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:55,354 : INFO : topic #1 (0.067): 0.022*\"people\" + 0.009*\"don\" + 0.006*\"make\" + 0.006*\"like\" + 0.006*\"know\" + 0.005*\"think\" + 0.005*\"just\" + 0.005*\"understand\" + 0.005*\"doesn\" + 0.004*\"really\"\n",
      "2017-06-05 20:56:55,356 : INFO : topic #7 (0.067): 0.009*\"war\" + 0.008*\"women\" + 0.005*\"backblaze\" + 0.004*\"did\" + 0.004*\"labour\" + 0.004*\"say\" + 0.004*\"like\" + 0.004*\"just\" + 0.004*\"years\" + 0.004*\"world\"\n",
      "2017-06-05 20:56:55,358 : INFO : topic #11 (0.067): 0.010*\"food\" + 0.008*\"http\" + 0.006*\"like\" + 0.005*\"article\" + 0.005*\"police\" + 0.005*\"think\" + 0.005*\"people\" + 0.004*\"www\" + 0.004*\"http www\" + 0.004*\"just\"\n",
      "2017-06-05 20:56:55,361 : INFO : topic #14 (0.067): 0.014*\"like\" + 0.013*\"data\" + 0.008*\"use\" + 0.005*\"java\" + 0.005*\"code\" + 0.005*\"just\" + 0.004*\"change\" + 0.004*\"things\" + 0.004*\"storage\" + 0.004*\"cases\"\n",
      "2017-06-05 20:56:55,363 : INFO : topic #6 (0.067): 0.015*\"just\" + 0.014*\"think\" + 0.013*\"people\" + 0.012*\"don\" + 0.009*\"like\" + 0.008*\"really\" + 0.007*\"work\" + 0.007*\"way\" + 0.007*\"need\" + 0.006*\"language\"\n",
      "2017-06-05 20:56:55,370 : INFO : topic diff=0.421542, rho=0.277350\n",
      "2017-06-05 20:56:55,394 : INFO : PROGRESS: pass 0, at document #28000/246175\n",
      "2017-06-05 20:56:57,121 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:56:57,900 : INFO : topic #13 (0.067): 0.008*\"just\" + 0.008*\"don\" + 0.008*\"things\" + 0.007*\"problem\" + 0.007*\"like\" + 0.006*\"time\" + 0.006*\"ai\" + 0.006*\"good\" + 0.005*\"google\" + 0.005*\"thanks\"\n",
      "2017-06-05 20:56:57,902 : INFO : topic #3 (0.067): 0.008*\"docker\" + 0.006*\"does\" + 0.006*\"don\" + 0.006*\"just\" + 0.006*\"work\" + 0.006*\"use\" + 0.005*\"want\" + 0.004*\"different\" + 0.004*\"need\" + 0.004*\"linux\"\n",
      "2017-06-05 20:56:57,905 : INFO : topic #4 (0.067): 0.009*\"people\" + 0.007*\"life\" + 0.006*\"like\" + 0.006*\"facebook\" + 0.005*\"don\" + 0.005*\"social\" + 0.004*\"think\" + 0.004*\"way\" + 0.004*\"time\" + 0.004*\"point\"\n",
      "2017-06-05 20:56:57,906 : INFO : topic #6 (0.067): 0.014*\"just\" + 0.014*\"think\" + 0.013*\"people\" + 0.012*\"don\" + 0.009*\"like\" + 0.008*\"really\" + 0.008*\"work\" + 0.007*\"programming\" + 0.007*\"way\" + 0.007*\"language\"\n",
      "2017-06-05 20:56:57,908 : INFO : topic #5 (0.067): 0.025*\"https\" + 0.014*\"org\" + 0.012*\"com\" + 0.010*\"github\" + 0.009*\"just\" + 0.008*\"web\" + 0.008*\"like\" + 0.007*\"using\" + 0.007*\"windows\" + 0.007*\"github com\"\n",
      "2017-06-05 20:56:57,915 : INFO : topic diff=0.383983, rho=0.267261\n",
      "2017-06-05 20:56:57,940 : INFO : PROGRESS: pass 0, at document #30000/246175\n",
      "2017-06-05 20:56:59,727 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:00,550 : INFO : topic #11 (0.067): 0.008*\"http\" + 0.007*\"food\" + 0.006*\"article\" + 0.006*\"like\" + 0.005*\"http www\" + 0.005*\"www\" + 0.004*\"police\" + 0.004*\"think\" + 0.004*\"people\" + 0.004*\"just\"\n",
      "2017-06-05 20:57:00,552 : INFO : topic #9 (0.067): 0.011*\"tax\" + 0.009*\"work\" + 0.008*\"pay\" + 0.007*\"year\" + 0.007*\"time\" + 0.006*\"news\" + 0.006*\"money\" + 0.006*\"people\" + 0.005*\"government\" + 0.005*\"don\"\n",
      "2017-06-05 20:57:00,553 : INFO : topic #12 (0.067): 0.031*\"com\" + 0.023*\"www\" + 0.021*\"https\" + 0.016*\"like\" + 0.015*\"https www\" + 0.012*\"http\" + 0.009*\"ago\" + 0.007*\"http www\" + 0.007*\"years\" + 0.006*\"time\"\n",
      "2017-06-05 20:57:00,555 : INFO : topic #2 (0.067): 0.009*\"make\" + 0.008*\"just\" + 0.008*\"google\" + 0.007*\"don\" + 0.007*\"time\" + 0.007*\"app\" + 0.007*\"react\" + 0.006*\"free\" + 0.006*\"money\" + 0.006*\"good\"\n",
      "2017-06-05 20:57:00,557 : INFO : topic #7 (0.067): 0.008*\"war\" + 0.006*\"labour\" + 0.005*\"women\" + 0.005*\"did\" + 0.004*\"just\" + 0.004*\"say\" + 0.004*\"like\" + 0.004*\"years\" + 0.003*\"think\" + 0.003*\"monero\"\n",
      "2017-06-05 20:57:00,564 : INFO : topic diff=0.402631, rho=0.258199\n",
      "2017-06-05 20:57:00,587 : INFO : PROGRESS: pass 0, at document #32000/246175\n",
      "2017-06-05 20:57:02,422 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:03,195 : INFO : topic #14 (0.067): 0.014*\"like\" + 0.012*\"data\" + 0.008*\"use\" + 0.006*\"games\" + 0.005*\"java\" + 0.004*\"just\" + 0.004*\"change\" + 0.004*\"used\" + 0.004*\"code\" + 0.004*\"don\"\n",
      "2017-06-05 20:57:03,197 : INFO : topic #8 (0.067): 0.008*\"like\" + 0.007*\"government\" + 0.007*\"airlines\" + 0.006*\"regulation\" + 0.006*\"people\" + 0.005*\"power\" + 0.005*\"bad\" + 0.005*\"think\" + 0.004*\"regulations\" + 0.004*\"just\"\n",
      "2017-06-05 20:57:03,199 : INFO : topic #7 (0.067): 0.009*\"openstack\" + 0.006*\"war\" + 0.005*\"did\" + 0.005*\"labour\" + 0.005*\"united\" + 0.004*\"just\" + 0.004*\"say\" + 0.004*\"women\" + 0.004*\"like\" + 0.004*\"years\"\n",
      "2017-06-05 20:57:03,201 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.014*\"org\" + 0.011*\"com\" + 0.010*\"github\" + 0.009*\"just\" + 0.008*\"wikipedia\" + 0.007*\"like\" + 0.007*\"en\" + 0.007*\"using\" + 0.007*\"wiki\"\n",
      "2017-06-05 20:57:03,203 : INFO : topic #4 (0.067): 0.008*\"people\" + 0.007*\"life\" + 0.006*\"like\" + 0.005*\"don\" + 0.005*\"social\" + 0.004*\"think\" + 0.004*\"facebook\" + 0.004*\"way\" + 0.004*\"time\" + 0.004*\"just\"\n",
      "2017-06-05 20:57:03,210 : INFO : topic diff=0.387487, rho=0.250000\n",
      "2017-06-05 20:57:03,243 : INFO : PROGRESS: pass 0, at document #34000/246175\n",
      "2017-06-05 20:57:05,249 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:06,026 : INFO : topic #2 (0.067): 0.009*\"make\" + 0.008*\"just\" + 0.008*\"don\" + 0.007*\"google\" + 0.007*\"time\" + 0.006*\"money\" + 0.006*\"facebook\" + 0.006*\"free\" + 0.006*\"good\" + 0.005*\"app\"\n",
      "2017-06-05 20:57:06,028 : INFO : topic #9 (0.067): 0.009*\"pay\" + 0.008*\"tax\" + 0.008*\"work\" + 0.007*\"money\" + 0.007*\"news\" + 0.006*\"time\" + 0.006*\"year\" + 0.006*\"people\" + 0.006*\"country\" + 0.005*\"don\"\n",
      "2017-06-05 20:57:06,030 : INFO : topic #3 (0.067): 0.007*\"does\" + 0.006*\"don\" + 0.006*\"use\" + 0.006*\"just\" + 0.006*\"work\" + 0.005*\"docker\" + 0.005*\"want\" + 0.005*\"different\" + 0.004*\"need\" + 0.004*\"ve\"\n",
      "2017-06-05 20:57:06,032 : INFO : topic #1 (0.067): 0.022*\"people\" + 0.010*\"don\" + 0.007*\"make\" + 0.007*\"like\" + 0.006*\"know\" + 0.006*\"think\" + 0.006*\"just\" + 0.005*\"doesn\" + 0.005*\"understand\" + 0.004*\"say\"\n",
      "2017-06-05 20:57:06,035 : INFO : topic #6 (0.067): 0.016*\"think\" + 0.015*\"just\" + 0.013*\"don\" + 0.012*\"people\" + 0.009*\"like\" + 0.009*\"really\" + 0.009*\"work\" + 0.008*\"way\" + 0.007*\"need\" + 0.006*\"time\"\n",
      "2017-06-05 20:57:06,042 : INFO : topic diff=0.394325, rho=0.242536\n",
      "2017-06-05 20:57:06,070 : INFO : PROGRESS: pass 0, at document #36000/246175\n",
      "2017-06-05 20:57:08,040 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:08,784 : INFO : topic #10 (0.067): 0.010*\"code\" + 0.010*\"use\" + 0.008*\"php\" + 0.006*\"just\" + 0.006*\"really\" + 0.005*\"memory\" + 0.004*\"want\" + 0.004*\"like\" + 0.004*\"way\" + 0.004*\"used\"\n",
      "2017-06-05 20:57:08,786 : INFO : topic #0 (0.067): 0.010*\"people\" + 0.008*\"airbnb\" + 0.006*\"twitter\" + 0.006*\"companies\" + 0.006*\"business\" + 0.006*\"hotel\" + 0.005*\"just\" + 0.004*\"don\" + 0.004*\"price\" + 0.004*\"point\"\n",
      "2017-06-05 20:57:08,789 : INFO : topic #2 (0.067): 0.009*\"make\" + 0.008*\"facebook\" + 0.008*\"just\" + 0.008*\"don\" + 0.007*\"google\" + 0.006*\"time\" + 0.006*\"money\" + 0.006*\"good\" + 0.006*\"free\" + 0.005*\"app\"\n",
      "2017-06-05 20:57:08,791 : INFO : topic #3 (0.067): 0.007*\"does\" + 0.006*\"don\" + 0.006*\"use\" + 0.006*\"just\" + 0.006*\"work\" + 0.006*\"want\" + 0.005*\"need\" + 0.004*\"different\" + 0.004*\"docker\" + 0.004*\"ve\"\n",
      "2017-06-05 20:57:08,793 : INFO : topic #8 (0.067): 0.008*\"like\" + 0.006*\"people\" + 0.006*\"government\" + 0.005*\"think\" + 0.005*\"power\" + 0.005*\"regulation\" + 0.005*\"bad\" + 0.004*\"airlines\" + 0.004*\"age\" + 0.004*\"good\"\n",
      "2017-06-05 20:57:08,802 : INFO : topic diff=0.384131, rho=0.235702\n",
      "2017-06-05 20:57:08,830 : INFO : PROGRESS: pass 0, at document #38000/246175\n",
      "2017-06-05 20:57:10,758 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:11,541 : INFO : topic #2 (0.067): 0.010*\"facebook\" + 0.009*\"make\" + 0.009*\"just\" + 0.008*\"don\" + 0.006*\"google\" + 0.006*\"users\" + 0.006*\"time\" + 0.006*\"money\" + 0.006*\"good\" + 0.005*\"use\"\n",
      "2017-06-05 20:57:11,543 : INFO : topic #1 (0.067): 0.023*\"people\" + 0.010*\"don\" + 0.007*\"like\" + 0.006*\"make\" + 0.006*\"know\" + 0.006*\"think\" + 0.006*\"just\" + 0.005*\"doesn\" + 0.005*\"understand\" + 0.005*\"say\"\n",
      "2017-06-05 20:57:11,545 : INFO : topic #8 (0.067): 0.008*\"like\" + 0.006*\"people\" + 0.005*\"power\" + 0.005*\"government\" + 0.005*\"think\" + 0.005*\"bad\" + 0.004*\"age\" + 0.004*\"regulation\" + 0.004*\"political\" + 0.004*\"good\"\n",
      "2017-06-05 20:57:11,548 : INFO : topic #14 (0.067): 0.015*\"like\" + 0.012*\"data\" + 0.009*\"use\" + 0.004*\"cases\" + 0.004*\"used\" + 0.004*\"just\" + 0.004*\"change\" + 0.004*\"latin\" + 0.004*\"games\" + 0.004*\"things\"\n",
      "2017-06-05 20:57:11,549 : INFO : topic #12 (0.067): 0.032*\"com\" + 0.025*\"www\" + 0.022*\"https\" + 0.016*\"https www\" + 0.015*\"like\" + 0.014*\"http\" + 0.009*\"http www\" + 0.008*\"ago\" + 0.007*\"years\" + 0.007*\"apple\"\n",
      "2017-06-05 20:57:11,556 : INFO : topic diff=0.353853, rho=0.229416\n",
      "2017-06-05 20:57:16,170 : INFO : -9.847 per-word bound, 920.7 perplexity estimate based on a held-out corpus of 2000 documents with 62852 words\n",
      "2017-06-05 20:57:16,171 : INFO : PROGRESS: pass 0, at document #40000/246175\n",
      "2017-06-05 20:57:18,004 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:18,746 : INFO : topic #4 (0.067): 0.009*\"people\" + 0.008*\"life\" + 0.006*\"like\" + 0.006*\"don\" + 0.006*\"facebook\" + 0.005*\"social\" + 0.005*\"think\" + 0.004*\"post\" + 0.004*\"hn\" + 0.004*\"way\"\n",
      "2017-06-05 20:57:18,748 : INFO : topic #9 (0.067): 0.017*\"tax\" + 0.010*\"money\" + 0.010*\"income\" + 0.010*\"pay\" + 0.010*\"taxes\" + 0.008*\"work\" + 0.007*\"news\" + 0.006*\"country\" + 0.006*\"countries\" + 0.006*\"rate\"\n",
      "2017-06-05 20:57:18,750 : INFO : topic #11 (0.067): 0.007*\"bull\" + 0.006*\"article\" + 0.006*\"http\" + 0.006*\"like\" + 0.005*\"food\" + 0.004*\"think\" + 0.004*\"http www\" + 0.004*\"just\" + 0.003*\"www\" + 0.003*\"radio\"\n",
      "2017-06-05 20:57:18,752 : INFO : topic #3 (0.067): 0.007*\"does\" + 0.007*\"use\" + 0.006*\"don\" + 0.006*\"just\" + 0.006*\"work\" + 0.005*\"want\" + 0.005*\"different\" + 0.005*\"need\" + 0.004*\"share\" + 0.004*\"ve\"\n",
      "2017-06-05 20:57:18,754 : INFO : topic #8 (0.067): 0.008*\"like\" + 0.006*\"people\" + 0.005*\"think\" + 0.005*\"government\" + 0.005*\"power\" + 0.005*\"age\" + 0.004*\"bad\" + 0.004*\"good\" + 0.004*\"political\" + 0.004*\"just\"\n",
      "2017-06-05 20:57:18,761 : INFO : topic diff=0.332383, rho=0.223607\n",
      "2017-06-05 20:57:18,788 : INFO : PROGRESS: pass 0, at document #42000/246175\n",
      "2017-06-05 20:57:20,683 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:21,381 : INFO : topic #6 (0.067): 0.015*\"think\" + 0.015*\"just\" + 0.013*\"don\" + 0.012*\"people\" + 0.010*\"like\" + 0.010*\"really\" + 0.009*\"work\" + 0.008*\"way\" + 0.007*\"need\" + 0.006*\"language\"\n",
      "2017-06-05 20:57:21,384 : INFO : topic #1 (0.067): 0.023*\"people\" + 0.009*\"don\" + 0.007*\"know\" + 0.007*\"like\" + 0.006*\"make\" + 0.006*\"think\" + 0.006*\"card\" + 0.006*\"just\" + 0.005*\"doesn\" + 0.005*\"say\"\n",
      "2017-06-05 20:57:21,386 : INFO : topic #4 (0.067): 0.009*\"people\" + 0.008*\"life\" + 0.006*\"like\" + 0.006*\"don\" + 0.005*\"social\" + 0.005*\"facebook\" + 0.004*\"think\" + 0.004*\"post\" + 0.004*\"hn\" + 0.004*\"way\"\n",
      "2017-06-05 20:57:21,388 : INFO : topic #3 (0.067): 0.007*\"use\" + 0.007*\"does\" + 0.006*\"just\" + 0.006*\"don\" + 0.006*\"work\" + 0.005*\"want\" + 0.005*\"different\" + 0.005*\"need\" + 0.004*\"ve\" + 0.004*\"make\"\n",
      "2017-06-05 20:57:21,390 : INFO : topic #10 (0.067): 0.011*\"code\" + 0.011*\"use\" + 0.006*\"just\" + 0.006*\"really\" + 0.005*\"memory\" + 0.004*\"php\" + 0.004*\"used\" + 0.004*\"like\" + 0.004*\"way\" + 0.004*\"js\"\n",
      "2017-06-05 20:57:21,397 : INFO : topic diff=0.337685, rho=0.218218\n",
      "2017-06-05 20:57:21,424 : INFO : PROGRESS: pass 0, at document #44000/246175\n",
      "2017-06-05 20:57:23,225 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:23,926 : INFO : topic #5 (0.067): 0.024*\"https\" + 0.015*\"org\" + 0.012*\"com\" + 0.010*\"windows\" + 0.010*\"en\" + 0.009*\"wikipedia\" + 0.009*\"web\" + 0.009*\"wiki\" + 0.008*\"org wiki\" + 0.008*\"wikipedia org\"\n",
      "2017-06-05 20:57:23,928 : INFO : topic #3 (0.067): 0.007*\"use\" + 0.006*\"does\" + 0.006*\"just\" + 0.006*\"don\" + 0.006*\"want\" + 0.005*\"work\" + 0.004*\"different\" + 0.004*\"need\" + 0.004*\"ve\" + 0.004*\"make\"\n",
      "2017-06-05 20:57:23,930 : INFO : topic #11 (0.067): 0.006*\"article\" + 0.006*\"like\" + 0.005*\"http\" + 0.005*\"food\" + 0.005*\"nz\" + 0.004*\"bull\" + 0.004*\"water\" + 0.004*\"police\" + 0.003*\"think\" + 0.003*\"just\"\n",
      "2017-06-05 20:57:23,933 : INFO : topic #14 (0.067): 0.015*\"like\" + 0.012*\"data\" + 0.009*\"use\" + 0.005*\"cases\" + 0.005*\"used\" + 0.004*\"change\" + 0.004*\"just\" + 0.004*\"code\" + 0.004*\"type\" + 0.004*\"latin\"\n",
      "2017-06-05 20:57:23,935 : INFO : topic #7 (0.067): 0.006*\"did\" + 0.005*\"war\" + 0.005*\"united\" + 0.005*\"russian\" + 0.004*\"say\" + 0.004*\"author\" + 0.004*\"reddit\" + 0.004*\"patent\" + 0.004*\"history\" + 0.004*\"world\"\n",
      "2017-06-05 20:57:23,942 : INFO : topic diff=0.323983, rho=0.213201\n",
      "2017-06-05 20:57:23,970 : INFO : PROGRESS: pass 0, at document #46000/246175\n",
      "2017-06-05 20:57:25,862 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:26,535 : INFO : topic #9 (0.067): 0.017*\"tax\" + 0.011*\"pay\" + 0.011*\"money\" + 0.009*\"taxes\" + 0.008*\"income\" + 0.008*\"work\" + 0.008*\"year\" + 0.006*\"news\" + 0.006*\"country\" + 0.006*\"people\"\n",
      "2017-06-05 20:57:26,537 : INFO : topic #0 (0.067): 0.009*\"people\" + 0.007*\"business\" + 0.007*\"companies\" + 0.005*\"uber\" + 0.005*\"just\" + 0.005*\"price\" + 0.004*\"point\" + 0.004*\"twitter\" + 0.004*\"cars\" + 0.004*\"airline\"\n",
      "2017-06-05 20:57:26,540 : INFO : topic #7 (0.067): 0.007*\"united\" + 0.006*\"did\" + 0.005*\"war\" + 0.004*\"reddit\" + 0.004*\"say\" + 0.004*\"russian\" + 0.004*\"history\" + 0.004*\"author\" + 0.004*\"just\" + 0.004*\"world\"\n",
      "2017-06-05 20:57:26,541 : INFO : topic #12 (0.067): 0.034*\"com\" + 0.026*\"www\" + 0.023*\"https\" + 0.018*\"apple\" + 0.016*\"https www\" + 0.014*\"http\" + 0.013*\"like\" + 0.009*\"http www\" + 0.009*\"ago\" + 0.007*\"years\"\n",
      "2017-06-05 20:57:26,543 : INFO : topic #10 (0.067): 0.012*\"code\" + 0.010*\"use\" + 0.007*\"memory\" + 0.006*\"just\" + 0.005*\"really\" + 0.004*\"used\" + 0.004*\"machine\" + 0.004*\"like\" + 0.004*\"easy\" + 0.004*\"want\"\n",
      "2017-06-05 20:57:26,550 : INFO : topic diff=0.323374, rho=0.208514\n",
      "2017-06-05 20:57:26,579 : INFO : PROGRESS: pass 0, at document #48000/246175\n",
      "2017-06-05 20:57:28,451 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:29,114 : INFO : topic #9 (0.067): 0.020*\"tax\" + 0.011*\"pay\" + 0.011*\"money\" + 0.010*\"taxes\" + 0.008*\"income\" + 0.008*\"year\" + 0.008*\"work\" + 0.006*\"news\" + 0.006*\"people\" + 0.005*\"country\"\n",
      "2017-06-05 20:57:29,117 : INFO : topic #10 (0.067): 0.013*\"code\" + 0.010*\"use\" + 0.008*\"memory\" + 0.005*\"just\" + 0.005*\"really\" + 0.004*\"machine\" + 0.004*\"like\" + 0.004*\"used\" + 0.004*\"time\" + 0.004*\"easy\"\n",
      "2017-06-05 20:57:29,119 : INFO : topic #5 (0.067): 0.024*\"https\" + 0.015*\"org\" + 0.012*\"com\" + 0.010*\"windows\" + 0.009*\"en\" + 0.009*\"wikipedia\" + 0.008*\"wiki\" + 0.008*\"github\" + 0.008*\"web\" + 0.008*\"org wiki\"\n",
      "2017-06-05 20:57:29,122 : INFO : topic #8 (0.067): 0.008*\"like\" + 0.006*\"people\" + 0.006*\"power\" + 0.005*\"government\" + 0.005*\"flight\" + 0.004*\"think\" + 0.004*\"bad\" + 0.004*\"good\" + 0.004*\"rights\" + 0.004*\"just\"\n",
      "2017-06-05 20:57:29,124 : INFO : topic #3 (0.067): 0.007*\"use\" + 0.007*\"just\" + 0.006*\"does\" + 0.006*\"don\" + 0.005*\"want\" + 0.005*\"work\" + 0.004*\"ve\" + 0.004*\"different\" + 0.004*\"need\" + 0.004*\"information\"\n",
      "2017-06-05 20:57:29,132 : INFO : topic diff=0.306811, rho=0.204124\n",
      "2017-06-05 20:57:29,159 : INFO : PROGRESS: pass 0, at document #50000/246175\n",
      "2017-06-05 20:57:30,984 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:31,616 : INFO : topic #14 (0.067): 0.015*\"like\" + 0.011*\"data\" + 0.009*\"use\" + 0.005*\"games\" + 0.005*\"cases\" + 0.005*\"used\" + 0.004*\"code\" + 0.004*\"looks\" + 0.004*\"just\" + 0.004*\"bit\"\n",
      "2017-06-05 20:57:31,617 : INFO : topic #0 (0.067): 0.009*\"people\" + 0.007*\"business\" + 0.006*\"companies\" + 0.005*\"just\" + 0.005*\"price\" + 0.004*\"point\" + 0.004*\"cars\" + 0.004*\"twitter\" + 0.004*\"uber\" + 0.004*\"tech\"\n",
      "2017-06-05 20:57:31,620 : INFO : topic #6 (0.067): 0.016*\"just\" + 0.015*\"think\" + 0.014*\"don\" + 0.013*\"people\" + 0.011*\"like\" + 0.010*\"really\" + 0.010*\"work\" + 0.009*\"way\" + 0.006*\"need\" + 0.006*\"time\"\n",
      "2017-06-05 20:57:31,622 : INFO : topic #2 (0.067): 0.010*\"make\" + 0.008*\"just\" + 0.008*\"google\" + 0.008*\"don\" + 0.007*\"ad\" + 0.006*\"users\" + 0.006*\"ads\" + 0.006*\"time\" + 0.006*\"use\" + 0.005*\"money\"\n",
      "2017-06-05 20:57:31,623 : INFO : topic #3 (0.067): 0.007*\"use\" + 0.007*\"just\" + 0.006*\"does\" + 0.006*\"don\" + 0.005*\"want\" + 0.005*\"work\" + 0.005*\"ve\" + 0.004*\"need\" + 0.004*\"desktop\" + 0.004*\"different\"\n",
      "2017-06-05 20:57:31,630 : INFO : topic diff=0.292780, rho=0.200000\n",
      "2017-06-05 20:57:31,659 : INFO : PROGRESS: pass 0, at document #52000/246175\n",
      "2017-06-05 20:57:33,525 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:34,157 : INFO : topic #2 (0.067): 0.010*\"make\" + 0.008*\"google\" + 0.008*\"just\" + 0.008*\"don\" + 0.007*\"content\" + 0.006*\"users\" + 0.006*\"money\" + 0.006*\"time\" + 0.006*\"ad\" + 0.006*\"free\"\n",
      "2017-06-05 20:57:34,160 : INFO : topic #4 (0.067): 0.013*\"life\" + 0.010*\"people\" + 0.007*\"snapchat\" + 0.007*\"social\" + 0.007*\"like\" + 0.005*\"don\" + 0.004*\"facebook\" + 0.004*\"laptop\" + 0.004*\"think\" + 0.004*\"way\"\n",
      "2017-06-05 20:57:34,162 : INFO : topic #14 (0.067): 0.016*\"like\" + 0.011*\"data\" + 0.009*\"use\" + 0.005*\"games\" + 0.005*\"cases\" + 0.004*\"used\" + 0.004*\"looks\" + 0.004*\"type\" + 0.004*\"just\" + 0.004*\"code\"\n",
      "2017-06-05 20:57:34,165 : INFO : topic #8 (0.067): 0.009*\"like\" + 0.006*\"people\" + 0.006*\"flight\" + 0.006*\"power\" + 0.005*\"airlines\" + 0.005*\"rights\" + 0.004*\"think\" + 0.004*\"bad\" + 0.004*\"united\" + 0.004*\"government\"\n",
      "2017-06-05 20:57:34,167 : INFO : topic #7 (0.067): 0.008*\"united\" + 0.007*\"did\" + 0.006*\"history\" + 0.006*\"reddit\" + 0.005*\"war\" + 0.004*\"say\" + 0.004*\"art\" + 0.004*\"just\" + 0.004*\"like\" + 0.004*\"years\"\n",
      "2017-06-05 20:57:34,175 : INFO : topic diff=0.296875, rho=0.196116\n",
      "2017-06-05 20:57:34,203 : INFO : PROGRESS: pass 0, at document #54000/246175\n",
      "2017-06-05 20:57:35,965 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:36,581 : INFO : topic #6 (0.067): 0.016*\"just\" + 0.015*\"think\" + 0.014*\"don\" + 0.013*\"people\" + 0.012*\"like\" + 0.011*\"work\" + 0.011*\"really\" + 0.008*\"way\" + 0.007*\"need\" + 0.006*\"time\"\n",
      "2017-06-05 20:57:36,583 : INFO : topic #1 (0.067): 0.024*\"people\" + 0.010*\"don\" + 0.007*\"like\" + 0.007*\"know\" + 0.006*\"think\" + 0.006*\"just\" + 0.006*\"make\" + 0.006*\"doesn\" + 0.005*\"card\" + 0.005*\"say\"\n",
      "2017-06-05 20:57:36,586 : INFO : topic #0 (0.067): 0.009*\"people\" + 0.007*\"business\" + 0.006*\"companies\" + 0.005*\"airline\" + 0.005*\"just\" + 0.005*\"price\" + 0.004*\"point\" + 0.004*\"ve\" + 0.004*\"uber\" + 0.004*\"cost\"\n",
      "2017-06-05 20:57:36,588 : INFO : topic #13 (0.067): 0.011*\"problem\" + 0.008*\"things\" + 0.008*\"time\" + 0.008*\"just\" + 0.007*\"good\" + 0.006*\"lot\" + 0.006*\"don\" + 0.006*\"like\" + 0.006*\"software\" + 0.005*\"real\"\n",
      "2017-06-05 20:57:36,590 : INFO : topic #11 (0.067): 0.009*\"police\" + 0.005*\"article\" + 0.005*\"like\" + 0.005*\"city\" + 0.005*\"energy\" + 0.005*\"water\" + 0.005*\"boarding\" + 0.004*\"food\" + 0.004*\"http\" + 0.004*\"nz\"\n",
      "2017-06-05 20:57:36,597 : INFO : topic diff=0.269220, rho=0.192450\n",
      "2017-06-05 20:57:36,624 : INFO : PROGRESS: pass 0, at document #56000/246175\n",
      "2017-06-05 20:57:38,381 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:38,981 : INFO : topic #10 (0.067): 0.013*\"code\" + 0.011*\"use\" + 0.007*\"memory\" + 0.005*\"just\" + 0.005*\"really\" + 0.004*\"js\" + 0.004*\"time\" + 0.004*\"like\" + 0.004*\"using\" + 0.004*\"used\"\n",
      "2017-06-05 20:57:38,983 : INFO : topic #2 (0.067): 0.012*\"google\" + 0.010*\"make\" + 0.009*\"don\" + 0.008*\"just\" + 0.006*\"users\" + 0.006*\"time\" + 0.006*\"content\" + 0.006*\"free\" + 0.006*\"use\" + 0.006*\"phone\"\n",
      "2017-06-05 20:57:38,985 : INFO : topic #11 (0.067): 0.010*\"police\" + 0.005*\"boarding\" + 0.005*\"article\" + 0.005*\"food\" + 0.005*\"like\" + 0.005*\"water\" + 0.004*\"city\" + 0.004*\"energy\" + 0.004*\"http\" + 0.003*\"just\"\n",
      "2017-06-05 20:57:38,988 : INFO : topic #13 (0.067): 0.011*\"problem\" + 0.008*\"things\" + 0.008*\"time\" + 0.008*\"just\" + 0.007*\"good\" + 0.006*\"don\" + 0.006*\"lot\" + 0.006*\"software\" + 0.006*\"like\" + 0.005*\"real\"\n",
      "2017-06-05 20:57:38,990 : INFO : topic #4 (0.067): 0.012*\"life\" + 0.010*\"people\" + 0.006*\"like\" + 0.005*\"don\" + 0.005*\"social\" + 0.005*\"snapchat\" + 0.004*\"way\" + 0.004*\"think\" + 0.004*\"laptop\" + 0.004*\"hn\"\n",
      "2017-06-05 20:57:38,996 : INFO : topic diff=0.258275, rho=0.188982\n",
      "2017-06-05 20:57:39,025 : INFO : PROGRESS: pass 0, at document #58000/246175\n",
      "2017-06-05 20:57:40,778 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:41,367 : INFO : topic #1 (0.067): 0.025*\"people\" + 0.011*\"don\" + 0.008*\"card\" + 0.007*\"know\" + 0.007*\"like\" + 0.006*\"think\" + 0.006*\"just\" + 0.006*\"make\" + 0.006*\"doesn\" + 0.005*\"say\"\n",
      "2017-06-05 20:57:41,370 : INFO : topic #10 (0.067): 0.013*\"code\" + 0.011*\"use\" + 0.008*\"memory\" + 0.005*\"just\" + 0.005*\"really\" + 0.005*\"time\" + 0.004*\"js\" + 0.004*\"like\" + 0.004*\"using\" + 0.004*\"performance\"\n",
      "2017-06-05 20:57:41,372 : INFO : topic #14 (0.067): 0.016*\"like\" + 0.012*\"data\" + 0.009*\"use\" + 0.006*\"type\" + 0.005*\"cases\" + 0.004*\"case\" + 0.004*\"used\" + 0.004*\"looks\" + 0.004*\"read\" + 0.004*\"just\"\n",
      "2017-06-05 20:57:41,375 : INFO : topic #0 (0.067): 0.010*\"people\" + 0.008*\"business\" + 0.006*\"companies\" + 0.006*\"airline\" + 0.005*\"just\" + 0.005*\"uber\" + 0.004*\"cost\" + 0.004*\"price\" + 0.004*\"ve\" + 0.004*\"point\"\n",
      "2017-06-05 20:57:41,377 : INFO : topic #12 (0.067): 0.038*\"com\" + 0.028*\"www\" + 0.026*\"https\" + 0.018*\"https www\" + 0.017*\"http\" + 0.013*\"like\" + 0.011*\"apple\" + 0.009*\"ago\" + 0.009*\"http www\" + 0.009*\"watch\"\n",
      "2017-06-05 20:57:41,385 : INFO : topic diff=0.244278, rho=0.185695\n",
      "2017-06-05 20:57:46,242 : INFO : -9.695 per-word bound, 829.1 perplexity estimate based on a held-out corpus of 2000 documents with 72122 words\n",
      "2017-06-05 20:57:46,243 : INFO : PROGRESS: pass 0, at document #60000/246175\n",
      "2017-06-05 20:57:48,076 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:48,652 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.011*\"don\" + 0.007*\"know\" + 0.007*\"card\" + 0.007*\"like\" + 0.006*\"doesn\" + 0.006*\"think\" + 0.006*\"just\" + 0.006*\"make\" + 0.006*\"understand\"\n",
      "2017-06-05 20:57:48,655 : INFO : topic #10 (0.067): 0.014*\"code\" + 0.011*\"use\" + 0.007*\"memory\" + 0.005*\"just\" + 0.005*\"really\" + 0.005*\"time\" + 0.004*\"like\" + 0.004*\"performance\" + 0.004*\"js\" + 0.004*\"using\"\n",
      "2017-06-05 20:57:48,657 : INFO : topic #8 (0.067): 0.008*\"like\" + 0.008*\"flight\" + 0.006*\"people\" + 0.005*\"united\" + 0.005*\"airlines\" + 0.005*\"power\" + 0.005*\"right\" + 0.005*\"bad\" + 0.004*\"think\" + 0.004*\"rights\"\n",
      "2017-06-05 20:57:48,660 : INFO : topic #0 (0.067): 0.010*\"people\" + 0.008*\"business\" + 0.006*\"companies\" + 0.006*\"airline\" + 0.005*\"just\" + 0.005*\"cost\" + 0.005*\"price\" + 0.004*\"ve\" + 0.004*\"uber\" + 0.004*\"point\"\n",
      "2017-06-05 20:57:48,662 : INFO : topic #7 (0.067): 0.011*\"united\" + 0.009*\"history\" + 0.007*\"did\" + 0.006*\"war\" + 0.004*\"china\" + 0.004*\"say\" + 0.004*\"reddit\" + 0.004*\"just\" + 0.003*\"lyft\" + 0.003*\"years\"\n",
      "2017-06-05 20:57:48,669 : INFO : topic diff=0.251142, rho=0.182574\n",
      "2017-06-05 20:57:48,697 : INFO : PROGRESS: pass 0, at document #62000/246175\n",
      "2017-06-05 20:57:50,435 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:50,990 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.010*\"people\" + 0.006*\"like\" + 0.006*\"social\" + 0.005*\"don\" + 0.004*\"point\" + 0.004*\"way\" + 0.004*\"think\" + 0.004*\"thing\" + 0.004*\"laptop\"\n",
      "2017-06-05 20:57:50,992 : INFO : topic #7 (0.067): 0.009*\"united\" + 0.008*\"history\" + 0.007*\"did\" + 0.006*\"war\" + 0.004*\"say\" + 0.004*\"reddit\" + 0.004*\"women\" + 0.004*\"china\" + 0.003*\"just\" + 0.003*\"like\"\n",
      "2017-06-05 20:57:50,994 : INFO : topic #14 (0.067): 0.016*\"like\" + 0.012*\"data\" + 0.009*\"use\" + 0.006*\"type\" + 0.005*\"cases\" + 0.004*\"used\" + 0.004*\"looks\" + 0.004*\"case\" + 0.004*\"read\" + 0.004*\"bit\"\n",
      "2017-06-05 20:57:50,996 : INFO : topic #2 (0.067): 0.014*\"google\" + 0.013*\"ads\" + 0.010*\"ad\" + 0.010*\"make\" + 0.009*\"don\" + 0.008*\"just\" + 0.008*\"open\" + 0.007*\"content\" + 0.007*\"users\" + 0.007*\"free\"\n",
      "2017-06-05 20:57:50,999 : INFO : topic #13 (0.067): 0.011*\"problem\" + 0.009*\"things\" + 0.008*\"time\" + 0.008*\"just\" + 0.007*\"good\" + 0.006*\"don\" + 0.006*\"software\" + 0.006*\"lot\" + 0.006*\"like\" + 0.005*\"thanks\"\n",
      "2017-06-05 20:57:51,005 : INFO : topic diff=0.233810, rho=0.179605\n",
      "2017-06-05 20:57:51,032 : INFO : PROGRESS: pass 0, at document #64000/246175\n",
      "2017-06-05 20:57:52,765 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:53,309 : INFO : topic #0 (0.067): 0.010*\"uber\" + 0.010*\"people\" + 0.008*\"business\" + 0.006*\"companies\" + 0.006*\"airline\" + 0.005*\"price\" + 0.005*\"just\" + 0.005*\"ve\" + 0.004*\"cost\" + 0.004*\"point\"\n",
      "2017-06-05 20:57:53,311 : INFO : topic #3 (0.067): 0.007*\"use\" + 0.007*\"linux\" + 0.007*\"just\" + 0.006*\"desktop\" + 0.006*\"don\" + 0.006*\"does\" + 0.006*\"want\" + 0.005*\"ve\" + 0.005*\"work\" + 0.004*\"user\"\n",
      "2017-06-05 20:57:53,314 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.011*\"don\" + 0.007*\"know\" + 0.007*\"like\" + 0.006*\"think\" + 0.006*\"doesn\" + 0.006*\"just\" + 0.006*\"make\" + 0.006*\"card\" + 0.005*\"say\"\n",
      "2017-06-05 20:57:53,316 : INFO : topic #5 (0.067): 0.024*\"https\" + 0.016*\"windows\" + 0.013*\"org\" + 0.011*\"com\" + 0.009*\"use\" + 0.009*\"github\" + 0.009*\"en\" + 0.008*\"wikipedia\" + 0.008*\"linux\" + 0.008*\"wiki\"\n",
      "2017-06-05 20:57:53,318 : INFO : topic #8 (0.067): 0.007*\"like\" + 0.007*\"flight\" + 0.007*\"people\" + 0.005*\"right\" + 0.005*\"airlines\" + 0.005*\"united\" + 0.005*\"bad\" + 0.004*\"power\" + 0.004*\"think\" + 0.004*\"government\"\n",
      "2017-06-05 20:57:53,325 : INFO : topic diff=0.224254, rho=0.176777\n",
      "2017-06-05 20:57:53,352 : INFO : PROGRESS: pass 0, at document #66000/246175\n",
      "2017-06-05 20:57:55,079 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:55,612 : INFO : topic #12 (0.067): 0.042*\"com\" + 0.029*\"https\" + 0.027*\"www\" + 0.020*\"http\" + 0.018*\"https www\" + 0.013*\"like\" + 0.011*\"ago\" + 0.009*\"years\" + 0.009*\"http www\" + 0.008*\"youtube\"\n",
      "2017-06-05 20:57:55,615 : INFO : topic #7 (0.067): 0.010*\"united\" + 0.007*\"did\" + 0.007*\"history\" + 0.004*\"war\" + 0.004*\"lyft\" + 0.004*\"say\" + 0.004*\"reddit\" + 0.004*\"comment\" + 0.004*\"just\" + 0.003*\"women\"\n",
      "2017-06-05 20:57:55,617 : INFO : topic #6 (0.067): 0.016*\"just\" + 0.016*\"think\" + 0.016*\"don\" + 0.014*\"people\" + 0.012*\"like\" + 0.011*\"really\" + 0.010*\"work\" + 0.009*\"way\" + 0.008*\"need\" + 0.006*\"time\"\n",
      "2017-06-05 20:57:55,619 : INFO : topic #3 (0.067): 0.008*\"desktop\" + 0.008*\"use\" + 0.007*\"just\" + 0.006*\"linux\" + 0.006*\"does\" + 0.006*\"don\" + 0.006*\"want\" + 0.005*\"ve\" + 0.005*\"work\" + 0.005*\"android\"\n",
      "2017-06-05 20:57:55,622 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.011*\"don\" + 0.007*\"know\" + 0.007*\"like\" + 0.006*\"think\" + 0.006*\"doesn\" + 0.006*\"just\" + 0.006*\"make\" + 0.005*\"card\" + 0.005*\"say\"\n",
      "2017-06-05 20:57:55,629 : INFO : topic diff=0.212453, rho=0.174078\n",
      "2017-06-05 20:57:55,656 : INFO : PROGRESS: pass 0, at document #68000/246175\n",
      "2017-06-05 20:57:57,359 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:57:57,874 : INFO : topic #8 (0.067): 0.009*\"flight\" + 0.007*\"like\" + 0.006*\"people\" + 0.006*\"plane\" + 0.006*\"united\" + 0.006*\"airlines\" + 0.005*\"right\" + 0.005*\"bad\" + 0.005*\"power\" + 0.004*\"government\"\n",
      "2017-06-05 20:57:57,876 : INFO : topic #7 (0.067): 0.011*\"united\" + 0.007*\"did\" + 0.007*\"history\" + 0.005*\"war\" + 0.004*\"reddit\" + 0.004*\"say\" + 0.003*\"lyft\" + 0.003*\"just\" + 0.003*\"comment\" + 0.003*\"time\"\n",
      "2017-06-05 20:57:57,879 : INFO : topic #2 (0.067): 0.014*\"ads\" + 0.013*\"google\" + 0.010*\"ad\" + 0.009*\"don\" + 0.009*\"open\" + 0.008*\"make\" + 0.008*\"content\" + 0.008*\"users\" + 0.008*\"just\" + 0.008*\"free\"\n",
      "2017-06-05 20:57:57,881 : INFO : topic #14 (0.067): 0.017*\"like\" + 0.013*\"irc\" + 0.011*\"data\" + 0.009*\"use\" + 0.008*\"sick\" + 0.006*\"type\" + 0.005*\"cases\" + 0.005*\"used\" + 0.004*\"case\" + 0.004*\"looks\"\n",
      "2017-06-05 20:57:57,884 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.011*\"don\" + 0.007*\"know\" + 0.007*\"like\" + 0.006*\"think\" + 0.006*\"doesn\" + 0.006*\"just\" + 0.006*\"make\" + 0.005*\"say\" + 0.005*\"mean\"\n",
      "2017-06-05 20:57:57,891 : INFO : topic diff=0.209971, rho=0.171499\n",
      "2017-06-05 20:57:57,918 : INFO : PROGRESS: pass 0, at document #70000/246175\n",
      "2017-06-05 20:57:59,632 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:00,138 : INFO : topic #14 (0.067): 0.017*\"like\" + 0.012*\"irc\" + 0.011*\"data\" + 0.009*\"use\" + 0.007*\"sick\" + 0.005*\"type\" + 0.005*\"used\" + 0.004*\"cases\" + 0.004*\"case\" + 0.004*\"looks\"\n",
      "2017-06-05 20:58:00,141 : INFO : topic #5 (0.067): 0.022*\"https\" + 0.013*\"apps\" + 0.013*\"windows\" + 0.012*\"org\" + 0.011*\"web\" + 0.011*\"electron\" + 0.010*\"com\" + 0.009*\"use\" + 0.008*\"github\" + 0.008*\"using\"\n",
      "2017-06-05 20:58:00,144 : INFO : topic #7 (0.067): 0.010*\"united\" + 0.007*\"did\" + 0.006*\"history\" + 0.005*\"reddit\" + 0.005*\"war\" + 0.004*\"say\" + 0.003*\"comment\" + 0.003*\"just\" + 0.003*\"macos\" + 0.003*\"time\"\n",
      "2017-06-05 20:58:00,146 : INFO : topic #9 (0.067): 0.013*\"pay\" + 0.012*\"work\" + 0.010*\"money\" + 0.008*\"news\" + 0.007*\"year\" + 0.006*\"home\" + 0.006*\"time\" + 0.006*\"people\" + 0.006*\"paid\" + 0.006*\"item\"\n",
      "2017-06-05 20:58:00,148 : INFO : topic #10 (0.067): 0.013*\"code\" + 0.012*\"use\" + 0.007*\"js\" + 0.006*\"memory\" + 0.005*\"just\" + 0.005*\"javascript\" + 0.005*\"performance\" + 0.005*\"really\" + 0.004*\"write\" + 0.004*\"like\"\n",
      "2017-06-05 20:58:00,155 : INFO : topic diff=0.196463, rho=0.169031\n",
      "2017-06-05 20:58:00,183 : INFO : PROGRESS: pass 0, at document #72000/246175\n",
      "2017-06-05 20:58:01,907 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:02,400 : INFO : topic #10 (0.067): 0.014*\"code\" + 0.012*\"use\" + 0.006*\"memory\" + 0.006*\"rust\" + 0.006*\"js\" + 0.005*\"just\" + 0.005*\"performance\" + 0.005*\"really\" + 0.004*\"write\" + 0.004*\"machine\"\n",
      "2017-06-05 20:58:02,402 : INFO : topic #2 (0.067): 0.013*\"google\" + 0.010*\"ads\" + 0.009*\"don\" + 0.009*\"make\" + 0.008*\"content\" + 0.008*\"open\" + 0.008*\"ad\" + 0.008*\"just\" + 0.008*\"users\" + 0.007*\"free\"\n",
      "2017-06-05 20:58:02,404 : INFO : topic #3 (0.067): 0.010*\"desktop\" + 0.008*\"use\" + 0.007*\"just\" + 0.006*\"does\" + 0.006*\"don\" + 0.006*\"want\" + 0.005*\"linux\" + 0.005*\"ve\" + 0.005*\"platform\" + 0.005*\"app\"\n",
      "2017-06-05 20:58:02,407 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.009*\"people\" + 0.006*\"like\" + 0.006*\"hn\" + 0.005*\"social\" + 0.005*\"don\" + 0.004*\"post\" + 0.004*\"laptop\" + 0.004*\"way\" + 0.004*\"battery\"\n",
      "2017-06-05 20:58:02,409 : INFO : topic #6 (0.067): 0.017*\"don\" + 0.016*\"just\" + 0.016*\"think\" + 0.013*\"people\" + 0.013*\"like\" + 0.012*\"really\" + 0.010*\"work\" + 0.008*\"way\" + 0.008*\"need\" + 0.007*\"time\"\n",
      "2017-06-05 20:58:02,416 : INFO : topic diff=0.226840, rho=0.166667\n",
      "2017-06-05 20:58:02,444 : INFO : PROGRESS: pass 0, at document #74000/246175\n",
      "2017-06-05 20:58:04,180 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:04,674 : INFO : topic #0 (0.067): 0.009*\"people\" + 0.008*\"twitter\" + 0.007*\"business\" + 0.007*\"companies\" + 0.006*\"cost\" + 0.005*\"uber\" + 0.005*\"price\" + 0.005*\"just\" + 0.005*\"airline\" + 0.005*\"ve\"\n",
      "2017-06-05 20:58:04,676 : INFO : topic #3 (0.067): 0.009*\"desktop\" + 0.008*\"use\" + 0.007*\"just\" + 0.006*\"don\" + 0.006*\"does\" + 0.006*\"want\" + 0.005*\"docker\" + 0.005*\"ve\" + 0.004*\"platform\" + 0.004*\"linux\"\n",
      "2017-06-05 20:58:04,679 : INFO : topic #6 (0.067): 0.017*\"don\" + 0.016*\"just\" + 0.016*\"think\" + 0.013*\"people\" + 0.013*\"like\" + 0.012*\"really\" + 0.011*\"work\" + 0.008*\"way\" + 0.008*\"need\" + 0.007*\"time\"\n",
      "2017-06-05 20:58:04,681 : INFO : topic #9 (0.067): 0.012*\"work\" + 0.012*\"pay\" + 0.010*\"money\" + 0.008*\"year\" + 0.007*\"news\" + 0.006*\"home\" + 0.006*\"time\" + 0.006*\"job\" + 0.006*\"hours\" + 0.006*\"people\"\n",
      "2017-06-05 20:58:04,684 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.009*\"people\" + 0.006*\"social\" + 0.006*\"like\" + 0.006*\"hn\" + 0.005*\"don\" + 0.005*\"post\" + 0.004*\"way\" + 0.004*\"point\" + 0.004*\"laptop\"\n",
      "2017-06-05 20:58:04,691 : INFO : topic diff=0.200956, rho=0.164399\n",
      "2017-06-05 20:58:04,717 : INFO : PROGRESS: pass 0, at document #76000/246175\n",
      "2017-06-05 20:58:06,365 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:06,846 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.010*\"people\" + 0.006*\"like\" + 0.006*\"social\" + 0.005*\"hn\" + 0.005*\"don\" + 0.005*\"laptop\" + 0.005*\"post\" + 0.004*\"way\" + 0.004*\"pro\"\n",
      "2017-06-05 20:58:06,848 : INFO : topic #3 (0.067): 0.008*\"use\" + 0.008*\"desktop\" + 0.007*\"just\" + 0.006*\"does\" + 0.006*\"don\" + 0.006*\"want\" + 0.005*\"ve\" + 0.005*\"docker\" + 0.004*\"work\" + 0.004*\"platform\"\n",
      "2017-06-05 20:58:06,850 : INFO : topic #11 (0.067): 0.011*\"police\" + 0.006*\"nuclear\" + 0.006*\"energy\" + 0.005*\"food\" + 0.004*\"crew\" + 0.004*\"water\" + 0.004*\"boarding\" + 0.004*\"like\" + 0.004*\"article\" + 0.004*\"solar\"\n",
      "2017-06-05 20:58:06,853 : INFO : topic #6 (0.067): 0.017*\"don\" + 0.017*\"just\" + 0.016*\"think\" + 0.013*\"people\" + 0.013*\"like\" + 0.012*\"really\" + 0.011*\"work\" + 0.008*\"need\" + 0.008*\"way\" + 0.007*\"time\"\n",
      "2017-06-05 20:58:06,855 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.013*\"org\" + 0.011*\"windows\" + 0.010*\"web\" + 0.010*\"com\" + 0.010*\"apps\" + 0.009*\"use\" + 0.009*\"github\" + 0.008*\"en\" + 0.008*\"wikipedia\"\n",
      "2017-06-05 20:58:06,862 : INFO : topic diff=0.187365, rho=0.162221\n",
      "2017-06-05 20:58:06,888 : INFO : PROGRESS: pass 0, at document #78000/246175\n",
      "2017-06-05 20:58:08,612 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:09,078 : INFO : topic #0 (0.067): 0.009*\"meat\" + 0.009*\"people\" + 0.007*\"twitter\" + 0.007*\"business\" + 0.007*\"companies\" + 0.006*\"cost\" + 0.005*\"airline\" + 0.005*\"price\" + 0.005*\"just\" + 0.005*\"ve\"\n",
      "2017-06-05 20:58:09,080 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.014*\"org\" + 0.010*\"com\" + 0.010*\"windows\" + 0.009*\"web\" + 0.009*\"en\" + 0.009*\"github\" + 0.009*\"use\" + 0.009*\"server\" + 0.009*\"wikipedia\"\n",
      "2017-06-05 20:58:09,082 : INFO : topic #12 (0.067): 0.044*\"com\" + 0.031*\"https\" + 0.030*\"www\" + 0.024*\"http\" + 0.018*\"https www\" + 0.012*\"like\" + 0.011*\"http www\" + 0.011*\"ago\" + 0.010*\"years\" + 0.007*\"apple\"\n",
      "2017-06-05 20:58:09,084 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.010*\"people\" + 0.006*\"like\" + 0.005*\"hn\" + 0.005*\"social\" + 0.005*\"don\" + 0.004*\"post\" + 0.004*\"way\" + 0.004*\"laptop\" + 0.004*\"point\"\n",
      "2017-06-05 20:58:09,086 : INFO : topic #11 (0.067): 0.010*\"police\" + 0.005*\"energy\" + 0.005*\"nuclear\" + 0.005*\"food\" + 0.005*\"water\" + 0.004*\"article\" + 0.004*\"eat\" + 0.004*\"like\" + 0.004*\"crew\" + 0.004*\"boarding\"\n",
      "2017-06-05 20:58:09,093 : INFO : topic diff=0.174578, rho=0.160128\n",
      "2017-06-05 20:58:13,985 : INFO : -9.683 per-word bound, 822.1 perplexity estimate based on a held-out corpus of 2000 documents with 76713 words\n",
      "2017-06-05 20:58:13,987 : INFO : PROGRESS: pass 0, at document #80000/246175\n",
      "2017-06-05 20:58:15,732 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:16,195 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.010*\"people\" + 0.006*\"like\" + 0.005*\"hn\" + 0.005*\"social\" + 0.005*\"post\" + 0.004*\"don\" + 0.004*\"way\" + 0.004*\"point\" + 0.004*\"think\"\n",
      "2017-06-05 20:58:16,197 : INFO : topic #2 (0.067): 0.012*\"google\" + 0.010*\"software\" + 0.009*\"make\" + 0.009*\"don\" + 0.009*\"open\" + 0.008*\"source\" + 0.008*\"free\" + 0.008*\"just\" + 0.007*\"users\" + 0.006*\"content\"\n",
      "2017-06-05 20:58:16,199 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.007*\"know\" + 0.006*\"doesn\" + 0.006*\"like\" + 0.006*\"think\" + 0.006*\"say\" + 0.006*\"just\" + 0.005*\"make\" + 0.005*\"mean\"\n",
      "2017-06-05 20:58:16,201 : INFO : topic #10 (0.067): 0.014*\"code\" + 0.012*\"use\" + 0.006*\"rust\" + 0.005*\"memory\" + 0.005*\"just\" + 0.005*\"using\" + 0.004*\"performance\" + 0.004*\"really\" + 0.004*\"like\" + 0.004*\"write\"\n",
      "2017-06-05 20:58:16,204 : INFO : topic #14 (0.067): 0.017*\"like\" + 0.012*\"data\" + 0.009*\"use\" + 0.005*\"semantic\" + 0.005*\"looks\" + 0.005*\"type\" + 0.005*\"irc\" + 0.005*\"change\" + 0.005*\"used\" + 0.004*\"cases\"\n",
      "2017-06-05 20:58:16,210 : INFO : topic diff=0.191621, rho=0.158114\n",
      "2017-06-05 20:58:16,239 : INFO : PROGRESS: pass 0, at document #82000/246175\n",
      "2017-06-05 20:58:17,970 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:18,433 : INFO : topic #3 (0.067): 0.008*\"use\" + 0.007*\"just\" + 0.007*\"desktop\" + 0.006*\"does\" + 0.005*\"want\" + 0.005*\"don\" + 0.004*\"ve\" + 0.004*\"work\" + 0.004*\"need\" + 0.004*\"different\"\n",
      "2017-06-05 20:58:18,435 : INFO : topic #2 (0.067): 0.011*\"google\" + 0.010*\"software\" + 0.009*\"open\" + 0.009*\"make\" + 0.009*\"don\" + 0.009*\"source\" + 0.008*\"free\" + 0.008*\"users\" + 0.007*\"just\" + 0.007*\"product\"\n",
      "2017-06-05 20:58:18,437 : INFO : topic #9 (0.067): 0.013*\"work\" + 0.012*\"pay\" + 0.010*\"money\" + 0.008*\"job\" + 0.007*\"news\" + 0.007*\"year\" + 0.006*\"jobs\" + 0.006*\"rate\" + 0.006*\"item\" + 0.006*\"time\"\n",
      "2017-06-05 20:58:18,440 : INFO : topic #5 (0.067): 0.024*\"https\" + 0.015*\"org\" + 0.011*\"windows\" + 0.010*\"com\" + 0.009*\"github\" + 0.009*\"use\" + 0.009*\"en\" + 0.009*\"wikipedia\" + 0.008*\"web\" + 0.008*\"server\"\n",
      "2017-06-05 20:58:18,442 : INFO : topic #0 (0.067): 0.008*\"people\" + 0.007*\"companies\" + 0.007*\"business\" + 0.006*\"meat\" + 0.006*\"cost\" + 0.006*\"twitter\" + 0.005*\"price\" + 0.005*\"market\" + 0.005*\"uber\" + 0.005*\"just\"\n",
      "2017-06-05 20:58:18,448 : INFO : topic diff=0.181305, rho=0.156174\n",
      "2017-06-05 20:58:18,475 : INFO : PROGRESS: pass 0, at document #84000/246175\n",
      "2017-06-05 20:58:20,337 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:20,786 : INFO : topic #7 (0.067): 0.013*\"mysql\" + 0.007*\"did\" + 0.006*\"united\" + 0.005*\"history\" + 0.005*\"russian\" + 0.004*\"reddit\" + 0.004*\"war\" + 0.004*\"say\" + 0.003*\"italy\" + 0.003*\"just\"\n",
      "2017-06-05 20:58:20,788 : INFO : topic #2 (0.067): 0.010*\"google\" + 0.009*\"users\" + 0.009*\"open\" + 0.009*\"make\" + 0.009*\"software\" + 0.009*\"don\" + 0.008*\"source\" + 0.008*\"free\" + 0.008*\"product\" + 0.007*\"just\"\n",
      "2017-06-05 20:58:20,790 : INFO : topic #3 (0.067): 0.009*\"use\" + 0.007*\"just\" + 0.007*\"desktop\" + 0.006*\"does\" + 0.005*\"want\" + 0.005*\"don\" + 0.004*\"ve\" + 0.004*\"work\" + 0.004*\"need\" + 0.004*\"different\"\n",
      "2017-06-05 20:58:20,792 : INFO : topic #13 (0.067): 0.013*\"problem\" + 0.010*\"time\" + 0.009*\"good\" + 0.009*\"things\" + 0.008*\"just\" + 0.007*\"software\" + 0.007*\"lot\" + 0.006*\"better\" + 0.006*\"real\" + 0.006*\"like\"\n",
      "2017-06-05 20:58:20,795 : INFO : topic #1 (0.067): 0.024*\"people\" + 0.012*\"don\" + 0.007*\"know\" + 0.006*\"like\" + 0.006*\"doesn\" + 0.006*\"think\" + 0.006*\"say\" + 0.006*\"just\" + 0.005*\"make\" + 0.005*\"point\"\n",
      "2017-06-05 20:58:20,801 : INFO : topic diff=0.159369, rho=0.154303\n",
      "2017-06-05 20:58:20,828 : INFO : PROGRESS: pass 0, at document #86000/246175\n",
      "2017-06-05 20:58:22,482 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:22,925 : INFO : topic #10 (0.067): 0.016*\"code\" + 0.012*\"use\" + 0.005*\"using\" + 0.005*\"python\" + 0.005*\"just\" + 0.005*\"performance\" + 0.005*\"rust\" + 0.005*\"memory\" + 0.004*\"write\" + 0.004*\"language\"\n",
      "2017-06-05 20:58:22,928 : INFO : topic #9 (0.067): 0.014*\"pay\" + 0.013*\"work\" + 0.011*\"money\" + 0.007*\"job\" + 0.007*\"news\" + 0.007*\"hours\" + 0.006*\"year\" + 0.006*\"time\" + 0.006*\"item\" + 0.006*\"people\"\n",
      "2017-06-05 20:58:22,930 : INFO : topic #7 (0.067): 0.012*\"mysql\" + 0.007*\"did\" + 0.006*\"russian\" + 0.005*\"history\" + 0.005*\"united\" + 0.004*\"women\" + 0.004*\"russia\" + 0.004*\"war\" + 0.004*\"reddit\" + 0.004*\"consciousness\"\n",
      "2017-06-05 20:58:22,933 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.009*\"people\" + 0.006*\"like\" + 0.005*\"hn\" + 0.005*\"post\" + 0.004*\"social\" + 0.004*\"don\" + 0.004*\"title\" + 0.004*\"way\" + 0.004*\"point\"\n",
      "2017-06-05 20:58:22,935 : INFO : topic #8 (0.067): 0.007*\"like\" + 0.007*\"people\" + 0.005*\"power\" + 0.005*\"law\" + 0.005*\"government\" + 0.005*\"right\" + 0.005*\"bad\" + 0.004*\"flight\" + 0.004*\"think\" + 0.004*\"airlines\"\n",
      "2017-06-05 20:58:22,942 : INFO : topic diff=0.148484, rho=0.152499\n",
      "2017-06-05 20:58:22,971 : INFO : PROGRESS: pass 0, at document #88000/246175\n",
      "2017-06-05 20:58:24,582 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:25,017 : INFO : topic #10 (0.067): 0.016*\"code\" + 0.013*\"use\" + 0.006*\"using\" + 0.005*\"python\" + 0.005*\"just\" + 0.005*\"performance\" + 0.005*\"language\" + 0.005*\"memory\" + 0.004*\"write\" + 0.004*\"js\"\n",
      "2017-06-05 20:58:25,019 : INFO : topic #1 (0.067): 0.024*\"people\" + 0.012*\"don\" + 0.008*\"know\" + 0.006*\"like\" + 0.006*\"doesn\" + 0.006*\"think\" + 0.006*\"say\" + 0.006*\"just\" + 0.005*\"mean\" + 0.005*\"make\"\n",
      "2017-06-05 20:58:25,021 : INFO : topic #5 (0.067): 0.024*\"https\" + 0.014*\"org\" + 0.011*\"use\" + 0.010*\"web\" + 0.010*\"com\" + 0.010*\"windows\" + 0.009*\"github\" + 0.008*\"wikipedia\" + 0.008*\"wiki\" + 0.008*\"using\"\n",
      "2017-06-05 20:58:25,024 : INFO : topic #0 (0.067): 0.008*\"people\" + 0.007*\"business\" + 0.007*\"uber\" + 0.006*\"companies\" + 0.006*\"cost\" + 0.005*\"twitter\" + 0.005*\"parking\" + 0.005*\"price\" + 0.005*\"just\" + 0.005*\"cars\"\n",
      "2017-06-05 20:58:25,026 : INFO : topic #13 (0.067): 0.012*\"problem\" + 0.010*\"time\" + 0.009*\"things\" + 0.009*\"good\" + 0.008*\"just\" + 0.007*\"lot\" + 0.006*\"software\" + 0.006*\"better\" + 0.006*\"like\" + 0.006*\"real\"\n",
      "2017-06-05 20:58:25,033 : INFO : topic diff=0.149680, rho=0.150756\n",
      "2017-06-05 20:58:25,060 : INFO : PROGRESS: pass 0, at document #90000/246175\n",
      "2017-06-05 20:58:26,702 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:27,128 : INFO : topic #1 (0.067): 0.024*\"people\" + 0.012*\"don\" + 0.008*\"know\" + 0.006*\"like\" + 0.006*\"think\" + 0.006*\"doesn\" + 0.006*\"say\" + 0.006*\"just\" + 0.005*\"mean\" + 0.005*\"make\"\n",
      "2017-06-05 20:58:27,131 : INFO : topic #3 (0.067): 0.010*\"use\" + 0.007*\"just\" + 0.007*\"desktop\" + 0.006*\"does\" + 0.006*\"want\" + 0.005*\"don\" + 0.005*\"information\" + 0.004*\"ve\" + 0.004*\"work\" + 0.004*\"need\"\n",
      "2017-06-05 20:58:27,133 : INFO : topic #12 (0.067): 0.044*\"com\" + 0.032*\"https\" + 0.032*\"www\" + 0.023*\"http\" + 0.019*\"https www\" + 0.012*\"http www\" + 0.011*\"ago\" + 0.011*\"like\" + 0.010*\"years\" + 0.007*\"years ago\"\n",
      "2017-06-05 20:58:27,135 : INFO : topic #14 (0.067): 0.017*\"like\" + 0.016*\"data\" + 0.010*\"use\" + 0.006*\"type\" + 0.005*\"cases\" + 0.005*\"used\" + 0.004*\"case\" + 0.004*\"looks\" + 0.004*\"bit\" + 0.004*\"read\"\n",
      "2017-06-05 20:58:27,138 : INFO : topic #10 (0.067): 0.015*\"code\" + 0.013*\"use\" + 0.006*\"using\" + 0.005*\"python\" + 0.005*\"language\" + 0.005*\"just\" + 0.005*\"memory\" + 0.005*\"performance\" + 0.004*\"rust\" + 0.004*\"write\"\n",
      "2017-06-05 20:58:27,144 : INFO : topic diff=0.148601, rho=0.149071\n",
      "2017-06-05 20:58:27,170 : INFO : PROGRESS: pass 0, at document #92000/246175\n",
      "2017-06-05 20:58:28,796 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:29,208 : INFO : topic #12 (0.067): 0.043*\"com\" + 0.032*\"https\" + 0.031*\"www\" + 0.023*\"http\" + 0.018*\"https www\" + 0.012*\"http www\" + 0.012*\"ago\" + 0.011*\"like\" + 0.010*\"years\" + 0.007*\"years ago\"\n",
      "2017-06-05 20:58:29,210 : INFO : topic #1 (0.067): 0.024*\"people\" + 0.012*\"don\" + 0.008*\"know\" + 0.007*\"think\" + 0.006*\"like\" + 0.006*\"doesn\" + 0.006*\"say\" + 0.006*\"just\" + 0.005*\"make\" + 0.005*\"mean\"\n",
      "2017-06-05 20:58:29,213 : INFO : topic #13 (0.067): 0.012*\"problem\" + 0.011*\"time\" + 0.009*\"things\" + 0.009*\"good\" + 0.008*\"just\" + 0.007*\"lot\" + 0.006*\"like\" + 0.006*\"better\" + 0.006*\"software\" + 0.006*\"real\"\n",
      "2017-06-05 20:58:29,214 : INFO : topic #3 (0.067): 0.009*\"use\" + 0.007*\"just\" + 0.007*\"desktop\" + 0.006*\"information\" + 0.006*\"want\" + 0.006*\"does\" + 0.005*\"don\" + 0.005*\"ve\" + 0.004*\"work\" + 0.004*\"user\"\n",
      "2017-06-05 20:58:29,217 : INFO : topic #5 (0.067): 0.024*\"https\" + 0.015*\"org\" + 0.011*\"use\" + 0.010*\"windows\" + 0.009*\"web\" + 0.009*\"com\" + 0.009*\"github\" + 0.009*\"wikipedia\" + 0.008*\"wiki\" + 0.008*\"en\"\n",
      "2017-06-05 20:58:29,224 : INFO : topic diff=0.145028, rho=0.147442\n",
      "2017-06-05 20:58:29,249 : INFO : PROGRESS: pass 0, at document #94000/246175\n",
      "2017-06-05 20:58:30,828 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:31,230 : INFO : topic #14 (0.067): 0.018*\"data\" + 0.017*\"like\" + 0.009*\"use\" + 0.006*\"type\" + 0.005*\"looks\" + 0.005*\"used\" + 0.005*\"cases\" + 0.005*\"case\" + 0.004*\"read\" + 0.004*\"bit\"\n",
      "2017-06-05 20:58:31,233 : INFO : topic #5 (0.067): 0.024*\"https\" + 0.015*\"org\" + 0.011*\"windows\" + 0.011*\"use\" + 0.009*\"com\" + 0.009*\"web\" + 0.009*\"github\" + 0.009*\"wikipedia\" + 0.009*\"react\" + 0.008*\"en\"\n",
      "2017-06-05 20:58:31,235 : INFO : topic #9 (0.067): 0.013*\"pay\" + 0.012*\"work\" + 0.010*\"money\" + 0.010*\"salary\" + 0.008*\"job\" + 0.008*\"news\" + 0.007*\"year\" + 0.006*\"time\" + 0.006*\"company\" + 0.006*\"people\"\n",
      "2017-06-05 20:58:31,237 : INFO : topic #11 (0.067): 0.010*\"bookmarks\" + 0.009*\"nuclear\" + 0.005*\"energy\" + 0.004*\"police\" + 0.004*\"article\" + 0.004*\"food\" + 0.004*\"power\" + 0.004*\"city\" + 0.004*\"like\" + 0.003*\"water\"\n",
      "2017-06-05 20:58:31,239 : INFO : topic #7 (0.067): 0.007*\"mysql\" + 0.007*\"did\" + 0.006*\"chinese\" + 0.006*\"history\" + 0.005*\"war\" + 0.004*\"china\" + 0.004*\"reddit\" + 0.004*\"say\" + 0.004*\"women\" + 0.004*\"russian\"\n",
      "2017-06-05 20:58:31,246 : INFO : topic diff=0.131607, rho=0.145865\n",
      "2017-06-05 20:58:31,272 : INFO : PROGRESS: pass 0, at document #96000/246175\n",
      "2017-06-05 20:58:32,909 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:33,301 : INFO : topic #3 (0.067): 0.009*\"use\" + 0.007*\"just\" + 0.007*\"desktop\" + 0.006*\"information\" + 0.006*\"want\" + 0.006*\"does\" + 0.005*\"user\" + 0.005*\"don\" + 0.005*\"account\" + 0.004*\"different\"\n",
      "2017-06-05 20:58:33,304 : INFO : topic #6 (0.067): 0.018*\"don\" + 0.016*\"just\" + 0.016*\"think\" + 0.013*\"like\" + 0.013*\"people\" + 0.012*\"really\" + 0.011*\"work\" + 0.009*\"need\" + 0.008*\"way\" + 0.007*\"time\"\n",
      "2017-06-05 20:58:33,307 : INFO : topic #11 (0.067): 0.009*\"nuclear\" + 0.008*\"bookmarks\" + 0.005*\"energy\" + 0.004*\"article\" + 0.004*\"police\" + 0.004*\"power\" + 0.004*\"like\" + 0.004*\"city\" + 0.004*\"food\" + 0.003*\"water\"\n",
      "2017-06-05 20:58:33,310 : INFO : topic #8 (0.067): 0.008*\"government\" + 0.007*\"like\" + 0.006*\"people\" + 0.006*\"power\" + 0.006*\"right\" + 0.006*\"law\" + 0.005*\"political\" + 0.005*\"trump\" + 0.005*\"bad\" + 0.004*\"free\"\n",
      "2017-06-05 20:58:33,312 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.009*\"people\" + 0.008*\"pro\" + 0.006*\"hn\" + 0.006*\"like\" + 0.005*\"post\" + 0.004*\"social\" + 0.004*\"title\" + 0.004*\"don\" + 0.004*\"laptop\"\n",
      "2017-06-05 20:58:33,319 : INFO : topic diff=0.132764, rho=0.144338\n",
      "2017-06-05 20:58:33,346 : INFO : PROGRESS: pass 0, at document #98000/246175\n",
      "2017-06-05 20:58:35,040 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:35,435 : INFO : topic #8 (0.067): 0.008*\"government\" + 0.007*\"like\" + 0.006*\"people\" + 0.006*\"right\" + 0.006*\"power\" + 0.005*\"law\" + 0.005*\"political\" + 0.005*\"trump\" + 0.004*\"bad\" + 0.004*\"free\"\n",
      "2017-06-05 20:58:35,437 : INFO : topic #14 (0.067): 0.021*\"data\" + 0.018*\"like\" + 0.009*\"use\" + 0.005*\"type\" + 0.005*\"looks\" + 0.005*\"used\" + 0.004*\"case\" + 0.004*\"read\" + 0.004*\"cases\" + 0.004*\"bit\"\n",
      "2017-06-05 20:58:35,440 : INFO : topic #0 (0.067): 0.014*\"uber\" + 0.011*\"twitter\" + 0.008*\"people\" + 0.008*\"business\" + 0.007*\"price\" + 0.006*\"companies\" + 0.005*\"cost\" + 0.005*\"market\" + 0.005*\"speech\" + 0.005*\"just\"\n",
      "2017-06-05 20:58:35,442 : INFO : topic #6 (0.067): 0.018*\"don\" + 0.017*\"just\" + 0.016*\"think\" + 0.013*\"like\" + 0.013*\"people\" + 0.012*\"really\" + 0.011*\"work\" + 0.009*\"need\" + 0.008*\"way\" + 0.007*\"time\"\n",
      "2017-06-05 20:58:35,444 : INFO : topic #13 (0.067): 0.012*\"problem\" + 0.010*\"time\" + 0.009*\"things\" + 0.009*\"good\" + 0.008*\"just\" + 0.007*\"lot\" + 0.006*\"better\" + 0.006*\"software\" + 0.006*\"like\" + 0.005*\"problems\"\n",
      "2017-06-05 20:58:35,450 : INFO : topic diff=0.134227, rho=0.142857\n",
      "2017-06-05 20:58:39,851 : INFO : -9.569 per-word bound, 759.5 perplexity estimate based on a held-out corpus of 2000 documents with 67982 words\n",
      "2017-06-05 20:58:39,852 : INFO : PROGRESS: pass 0, at document #100000/246175\n",
      "2017-06-05 20:58:41,549 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:41,931 : INFO : topic #6 (0.067): 0.018*\"don\" + 0.017*\"just\" + 0.016*\"think\" + 0.014*\"like\" + 0.013*\"people\" + 0.012*\"really\" + 0.011*\"work\" + 0.008*\"need\" + 0.008*\"way\" + 0.007*\"time\"\n",
      "2017-06-05 20:58:41,933 : INFO : topic #8 (0.067): 0.008*\"government\" + 0.007*\"like\" + 0.006*\"people\" + 0.006*\"right\" + 0.006*\"law\" + 0.006*\"power\" + 0.005*\"political\" + 0.005*\"bad\" + 0.004*\"trump\" + 0.004*\"state\"\n",
      "2017-06-05 20:58:41,935 : INFO : topic #5 (0.067): 0.022*\"https\" + 0.014*\"org\" + 0.011*\"web\" + 0.010*\"use\" + 0.010*\"windows\" + 0.009*\"com\" + 0.009*\"github\" + 0.008*\"wikipedia\" + 0.008*\"wiki\" + 0.008*\"en\"\n",
      "2017-06-05 20:58:41,938 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.009*\"people\" + 0.008*\"pro\" + 0.007*\"hn\" + 0.006*\"post\" + 0.006*\"like\" + 0.006*\"social\" + 0.004*\"laptop\" + 0.004*\"title\" + 0.004*\"don\"\n",
      "2017-06-05 20:58:41,940 : INFO : topic #10 (0.067): 0.014*\"code\" + 0.012*\"use\" + 0.005*\"using\" + 0.005*\"js\" + 0.005*\"just\" + 0.005*\"memory\" + 0.004*\"performance\" + 0.004*\"build\" + 0.004*\"write\" + 0.004*\"python\"\n",
      "2017-06-05 20:58:41,947 : INFO : topic diff=0.127421, rho=0.141421\n",
      "2017-06-05 20:58:41,974 : INFO : PROGRESS: pass 0, at document #102000/246175\n",
      "2017-06-05 20:58:43,605 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:43,994 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.014*\"org\" + 0.010*\"use\" + 0.010*\"web\" + 0.010*\"com\" + 0.010*\"windows\" + 0.009*\"github\" + 0.008*\"wikipedia\" + 0.008*\"en\" + 0.008*\"wiki\"\n",
      "2017-06-05 20:58:43,996 : INFO : topic #3 (0.067): 0.009*\"use\" + 0.007*\"just\" + 0.007*\"desktop\" + 0.006*\"android\" + 0.006*\"want\" + 0.006*\"does\" + 0.006*\"information\" + 0.005*\"user\" + 0.005*\"don\" + 0.005*\"account\"\n",
      "2017-06-05 20:58:43,999 : INFO : topic #0 (0.067): 0.013*\"uber\" + 0.013*\"twitter\" + 0.008*\"people\" + 0.007*\"price\" + 0.007*\"business\" + 0.006*\"companies\" + 0.005*\"cost\" + 0.005*\"car\" + 0.005*\"market\" + 0.005*\"just\"\n",
      "2017-06-05 20:58:44,001 : INFO : topic #12 (0.067): 0.044*\"com\" + 0.032*\"https\" + 0.031*\"www\" + 0.024*\"http\" + 0.018*\"https www\" + 0.013*\"http www\" + 0.011*\"like\" + 0.011*\"ago\" + 0.010*\"apple\" + 0.010*\"years\"\n",
      "2017-06-05 20:58:44,004 : INFO : topic #1 (0.067): 0.025*\"people\" + 0.012*\"don\" + 0.007*\"know\" + 0.007*\"think\" + 0.007*\"doesn\" + 0.007*\"like\" + 0.006*\"say\" + 0.006*\"just\" + 0.006*\"mean\" + 0.005*\"does\"\n",
      "2017-06-05 20:58:44,011 : INFO : topic diff=0.116791, rho=0.140028\n",
      "2017-06-05 20:58:44,040 : INFO : PROGRESS: pass 0, at document #104000/246175\n",
      "2017-06-05 20:58:45,654 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:46,028 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.008*\"people\" + 0.006*\"pro\" + 0.006*\"hn\" + 0.006*\"social\" + 0.006*\"like\" + 0.006*\"post\" + 0.004*\"laptop\" + 0.004*\"title\" + 0.004*\"way\"\n",
      "2017-06-05 20:58:46,030 : INFO : topic #2 (0.067): 0.016*\"google\" + 0.011*\"open\" + 0.010*\"source\" + 0.010*\"don\" + 0.009*\"make\" + 0.009*\"software\" + 0.008*\"users\" + 0.008*\"free\" + 0.007*\"just\" + 0.007*\"network\"\n",
      "2017-06-05 20:58:46,032 : INFO : topic #8 (0.067): 0.008*\"government\" + 0.007*\"like\" + 0.006*\"right\" + 0.006*\"people\" + 0.006*\"law\" + 0.005*\"power\" + 0.005*\"bad\" + 0.004*\"political\" + 0.004*\"think\" + 0.004*\"state\"\n",
      "2017-06-05 20:58:46,034 : INFO : topic #0 (0.067): 0.016*\"uber\" + 0.011*\"twitter\" + 0.008*\"people\" + 0.008*\"business\" + 0.007*\"price\" + 0.006*\"companies\" + 0.006*\"car\" + 0.005*\"driving\" + 0.005*\"cost\" + 0.005*\"market\"\n",
      "2017-06-05 20:58:46,037 : INFO : topic #6 (0.067): 0.018*\"don\" + 0.017*\"just\" + 0.016*\"think\" + 0.014*\"like\" + 0.012*\"people\" + 0.012*\"really\" + 0.011*\"work\" + 0.009*\"need\" + 0.008*\"way\" + 0.007*\"time\"\n",
      "2017-06-05 20:58:46,043 : INFO : topic diff=0.115058, rho=0.138675\n",
      "2017-06-05 20:58:46,069 : INFO : PROGRESS: pass 0, at document #106000/246175\n",
      "2017-06-05 20:58:47,660 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:48,028 : INFO : topic #12 (0.067): 0.045*\"com\" + 0.032*\"https\" + 0.031*\"www\" + 0.024*\"http\" + 0.018*\"https www\" + 0.012*\"http www\" + 0.012*\"ago\" + 0.011*\"like\" + 0.010*\"years\" + 0.008*\"apple\"\n",
      "2017-06-05 20:58:48,030 : INFO : topic #11 (0.067): 0.006*\"nuclear\" + 0.005*\"bookmarks\" + 0.004*\"energy\" + 0.004*\"space\" + 0.004*\"food\" + 0.004*\"power\" + 0.004*\"article\" + 0.004*\"city\" + 0.004*\"like\" + 0.004*\"water\"\n",
      "2017-06-05 20:58:48,033 : INFO : topic #2 (0.067): 0.015*\"google\" + 0.011*\"open\" + 0.010*\"source\" + 0.010*\"don\" + 0.010*\"make\" + 0.009*\"software\" + 0.009*\"users\" + 0.008*\"free\" + 0.007*\"just\" + 0.007*\"product\"\n",
      "2017-06-05 20:58:48,036 : INFO : topic #7 (0.067): 0.007*\"did\" + 0.007*\"war\" + 0.007*\"history\" + 0.004*\"say\" + 0.004*\"mysql\" + 0.004*\"wasn\" + 0.004*\"origin\" + 0.004*\"lyft\" + 0.004*\"reddit\" + 0.004*\"world\"\n",
      "2017-06-05 20:58:48,038 : INFO : topic #14 (0.067): 0.021*\"data\" + 0.018*\"like\" + 0.008*\"use\" + 0.006*\"prolog\" + 0.005*\"looks\" + 0.005*\"case\" + 0.005*\"cases\" + 0.005*\"type\" + 0.005*\"used\" + 0.004*\"bit\"\n",
      "2017-06-05 20:58:48,046 : INFO : topic diff=0.112399, rho=0.137361\n",
      "2017-06-05 20:58:48,073 : INFO : PROGRESS: pass 0, at document #108000/246175\n",
      "2017-06-05 20:58:49,722 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:50,087 : INFO : topic #6 (0.067): 0.018*\"don\" + 0.017*\"just\" + 0.016*\"think\" + 0.014*\"like\" + 0.013*\"really\" + 0.012*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.008*\"way\" + 0.007*\"know\"\n",
      "2017-06-05 20:58:50,090 : INFO : topic #0 (0.067): 0.015*\"uber\" + 0.010*\"twitter\" + 0.008*\"people\" + 0.008*\"business\" + 0.007*\"price\" + 0.006*\"companies\" + 0.005*\"car\" + 0.005*\"cost\" + 0.005*\"market\" + 0.005*\"just\"\n",
      "2017-06-05 20:58:50,092 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.008*\"people\" + 0.007*\"pro\" + 0.006*\"post\" + 0.006*\"like\" + 0.006*\"hn\" + 0.006*\"social\" + 0.005*\"movie\" + 0.005*\"laptop\" + 0.004*\"title\"\n",
      "2017-06-05 20:58:50,094 : INFO : topic #7 (0.067): 0.007*\"did\" + 0.007*\"war\" + 0.007*\"history\" + 0.004*\"say\" + 0.004*\"reddit\" + 0.004*\"wasn\" + 0.004*\"origin\" + 0.003*\"just\" + 0.003*\"lyft\" + 0.003*\"world\"\n",
      "2017-06-05 20:58:50,096 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.015*\"org\" + 0.011*\"windows\" + 0.010*\"use\" + 0.010*\"com\" + 0.009*\"web\" + 0.009*\"github\" + 0.008*\"en\" + 0.008*\"wikipedia\" + 0.008*\"wiki\"\n",
      "2017-06-05 20:58:50,103 : INFO : topic diff=0.111235, rho=0.136083\n",
      "2017-06-05 20:58:50,130 : INFO : PROGRESS: pass 0, at document #110000/246175\n",
      "2017-06-05 20:58:51,785 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:52,146 : INFO : topic #7 (0.067): 0.007*\"did\" + 0.006*\"war\" + 0.006*\"history\" + 0.004*\"say\" + 0.004*\"reddit\" + 0.004*\"wasn\" + 0.004*\"author\" + 0.003*\"just\" + 0.003*\"origin\" + 0.003*\"world\"\n",
      "2017-06-05 20:58:52,148 : INFO : topic #8 (0.067): 0.007*\"government\" + 0.007*\"like\" + 0.006*\"power\" + 0.006*\"right\" + 0.006*\"people\" + 0.005*\"law\" + 0.005*\"coal\" + 0.005*\"political\" + 0.005*\"bad\" + 0.004*\"public\"\n",
      "2017-06-05 20:58:52,150 : INFO : topic #6 (0.067): 0.018*\"don\" + 0.017*\"just\" + 0.016*\"think\" + 0.014*\"like\" + 0.013*\"really\" + 0.012*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.007*\"way\" + 0.007*\"time\"\n",
      "2017-06-05 20:58:52,153 : INFO : topic #13 (0.067): 0.013*\"problem\" + 0.010*\"time\" + 0.009*\"good\" + 0.009*\"things\" + 0.008*\"just\" + 0.006*\"lot\" + 0.006*\"like\" + 0.006*\"real\" + 0.006*\"better\" + 0.006*\"problems\"\n",
      "2017-06-05 20:58:52,156 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.008*\"people\" + 0.007*\"post\" + 0.006*\"movie\" + 0.006*\"like\" + 0.006*\"pro\" + 0.006*\"hn\" + 0.006*\"social\" + 0.004*\"laptop\" + 0.004*\"friends\"\n",
      "2017-06-05 20:58:52,163 : INFO : topic diff=0.108332, rho=0.134840\n",
      "2017-06-05 20:58:52,192 : INFO : PROGRESS: pass 0, at document #112000/246175\n",
      "2017-06-05 20:58:53,907 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:54,273 : INFO : topic #0 (0.067): 0.013*\"uber\" + 0.009*\"twitter\" + 0.008*\"people\" + 0.007*\"business\" + 0.007*\"price\" + 0.006*\"companies\" + 0.006*\"car\" + 0.006*\"cost\" + 0.005*\"cars\" + 0.005*\"market\"\n",
      "2017-06-05 20:58:54,275 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.014*\"org\" + 0.011*\"windows\" + 0.010*\"com\" + 0.010*\"github\" + 0.010*\"use\" + 0.009*\"web\" + 0.008*\"en\" + 0.008*\"wikipedia\" + 0.007*\"wiki\"\n",
      "2017-06-05 20:58:54,278 : INFO : topic #6 (0.067): 0.018*\"don\" + 0.017*\"just\" + 0.016*\"think\" + 0.014*\"like\" + 0.013*\"really\" + 0.012*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.007*\"way\" + 0.007*\"time\"\n",
      "2017-06-05 20:58:54,280 : INFO : topic #9 (0.067): 0.012*\"pay\" + 0.012*\"money\" + 0.010*\"work\" + 0.008*\"news\" + 0.008*\"job\" + 0.007*\"year\" + 0.007*\"company\" + 0.007*\"id\" + 0.006*\"item\" + 0.006*\"time\"\n",
      "2017-06-05 20:58:54,282 : INFO : topic #12 (0.067): 0.047*\"com\" + 0.034*\"https\" + 0.031*\"www\" + 0.025*\"http\" + 0.018*\"https www\" + 0.012*\"http www\" + 0.011*\"ago\" + 0.011*\"like\" + 0.010*\"years\" + 0.009*\"apple\"\n",
      "2017-06-05 20:58:54,288 : INFO : topic diff=0.110168, rho=0.133631\n",
      "2017-06-05 20:58:54,316 : INFO : PROGRESS: pass 0, at document #114000/246175\n",
      "2017-06-05 20:58:56,009 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:56,357 : INFO : topic #0 (0.067): 0.011*\"uber\" + 0.008*\"people\" + 0.008*\"twitter\" + 0.007*\"business\" + 0.007*\"price\" + 0.006*\"companies\" + 0.006*\"cost\" + 0.006*\"car\" + 0.005*\"cars\" + 0.005*\"market\"\n",
      "2017-06-05 20:58:56,358 : INFO : topic #12 (0.067): 0.048*\"com\" + 0.035*\"https\" + 0.031*\"www\" + 0.025*\"http\" + 0.018*\"https www\" + 0.012*\"http www\" + 0.011*\"ago\" + 0.011*\"like\" + 0.010*\"years\" + 0.010*\"apple\"\n",
      "2017-06-05 20:58:56,361 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.009*\"pro\" + 0.008*\"people\" + 0.006*\"post\" + 0.006*\"like\" + 0.006*\"social\" + 0.006*\"hn\" + 0.005*\"movie\" + 0.005*\"laptop\" + 0.004*\"title\"\n",
      "2017-06-05 20:58:56,363 : INFO : topic #10 (0.067): 0.016*\"code\" + 0.012*\"use\" + 0.006*\"using\" + 0.005*\"build\" + 0.005*\"js\" + 0.005*\"python\" + 0.005*\"write\" + 0.005*\"just\" + 0.004*\"time\" + 0.004*\"design\"\n",
      "2017-06-05 20:58:56,365 : INFO : topic #8 (0.067): 0.007*\"like\" + 0.007*\"government\" + 0.006*\"political\" + 0.006*\"right\" + 0.006*\"power\" + 0.006*\"people\" + 0.005*\"bad\" + 0.005*\"law\" + 0.004*\"coal\" + 0.004*\"public\"\n",
      "2017-06-05 20:58:56,371 : INFO : topic diff=0.108806, rho=0.132453\n",
      "2017-06-05 20:58:56,400 : INFO : PROGRESS: pass 0, at document #116000/246175\n",
      "2017-06-05 20:58:58,022 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:58:58,367 : INFO : topic #6 (0.067): 0.018*\"don\" + 0.017*\"just\" + 0.016*\"think\" + 0.014*\"like\" + 0.014*\"really\" + 0.012*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.008*\"way\" + 0.007*\"time\"\n",
      "2017-06-05 20:58:58,370 : INFO : topic #14 (0.067): 0.025*\"data\" + 0.018*\"like\" + 0.008*\"use\" + 0.005*\"looks\" + 0.005*\"case\" + 0.005*\"type\" + 0.005*\"cases\" + 0.005*\"used\" + 0.005*\"change\" + 0.004*\"read\"\n",
      "2017-06-05 20:58:58,372 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.014*\"windows\" + 0.013*\"org\" + 0.010*\"github\" + 0.010*\"com\" + 0.010*\"use\" + 0.010*\"web\" + 0.008*\"en\" + 0.008*\"wikipedia\" + 0.007*\"wiki\"\n",
      "2017-06-05 20:58:58,375 : INFO : topic #8 (0.067): 0.007*\"like\" + 0.006*\"government\" + 0.006*\"right\" + 0.006*\"people\" + 0.006*\"political\" + 0.006*\"power\" + 0.005*\"bad\" + 0.005*\"law\" + 0.004*\"coal\" + 0.004*\"think\"\n",
      "2017-06-05 20:58:58,377 : INFO : topic #13 (0.067): 0.014*\"problem\" + 0.011*\"time\" + 0.010*\"good\" + 0.009*\"things\" + 0.008*\"just\" + 0.006*\"lot\" + 0.006*\"like\" + 0.006*\"ve\" + 0.006*\"real\" + 0.005*\"game\"\n",
      "2017-06-05 20:58:58,384 : INFO : topic diff=0.103069, rho=0.131306\n",
      "2017-06-05 20:58:58,412 : INFO : PROGRESS: pass 0, at document #118000/246175\n",
      "2017-06-05 20:59:00,039 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:00,382 : INFO : topic #7 (0.067): 0.007*\"did\" + 0.006*\"history\" + 0.006*\"gun\" + 0.005*\"reddit\" + 0.005*\"war\" + 0.005*\"macos\" + 0.004*\"notes\" + 0.004*\"say\" + 0.004*\"author\" + 0.004*\"wasn\"\n",
      "2017-06-05 20:59:00,385 : INFO : topic #1 (0.067): 0.025*\"people\" + 0.012*\"don\" + 0.007*\"know\" + 0.007*\"doesn\" + 0.007*\"think\" + 0.007*\"like\" + 0.006*\"say\" + 0.006*\"just\" + 0.006*\"make\" + 0.006*\"mean\"\n",
      "2017-06-05 20:59:00,387 : INFO : topic #10 (0.067): 0.016*\"code\" + 0.012*\"use\" + 0.005*\"using\" + 0.005*\"build\" + 0.005*\"python\" + 0.004*\"just\" + 0.004*\"js\" + 0.004*\"machine\" + 0.004*\"write\" + 0.004*\"time\"\n",
      "2017-06-05 20:59:00,390 : INFO : topic #9 (0.067): 0.014*\"money\" + 0.011*\"work\" + 0.011*\"pay\" + 0.009*\"job\" + 0.007*\"year\" + 0.007*\"company\" + 0.007*\"jobs\" + 0.007*\"people\" + 0.006*\"news\" + 0.006*\"time\"\n",
      "2017-06-05 20:59:00,392 : INFO : topic #11 (0.067): 0.006*\"city\" + 0.006*\"nuclear\" + 0.006*\"energy\" + 0.004*\"global\" + 0.004*\"power\" + 0.004*\"water\" + 0.004*\"food\" + 0.004*\"cities\" + 0.003*\"like\" + 0.003*\"space\"\n",
      "2017-06-05 20:59:00,399 : INFO : topic diff=0.113752, rho=0.130189\n",
      "2017-06-05 20:59:05,287 : INFO : -9.530 per-word bound, 739.3 perplexity estimate based on a held-out corpus of 2000 documents with 82011 words\n",
      "2017-06-05 20:59:05,288 : INFO : PROGRESS: pass 0, at document #120000/246175\n",
      "2017-06-05 20:59:06,955 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:07,293 : INFO : topic #0 (0.067): 0.008*\"car\" + 0.008*\"people\" + 0.007*\"business\" + 0.007*\"uber\" + 0.007*\"companies\" + 0.006*\"market\" + 0.006*\"price\" + 0.006*\"company\" + 0.006*\"cars\" + 0.005*\"cost\"\n",
      "2017-06-05 20:59:07,295 : INFO : topic #12 (0.067): 0.055*\"com\" + 0.037*\"https\" + 0.032*\"www\" + 0.026*\"http\" + 0.019*\"https www\" + 0.012*\"http www\" + 0.011*\"apple\" + 0.010*\"ago\" + 0.010*\"years\" + 0.010*\"like\"\n",
      "2017-06-05 20:59:07,298 : INFO : topic #5 (0.067): 0.024*\"https\" + 0.013*\"windows\" + 0.013*\"org\" + 0.011*\"com\" + 0.010*\"web\" + 0.010*\"github\" + 0.010*\"use\" + 0.008*\"en\" + 0.008*\"wikipedia\" + 0.007*\"wiki\"\n",
      "2017-06-05 20:59:07,300 : INFO : topic #3 (0.067): 0.009*\"use\" + 0.009*\"android\" + 0.007*\"desktop\" + 0.007*\"just\" + 0.006*\"want\" + 0.006*\"does\" + 0.006*\"user\" + 0.006*\"linux\" + 0.005*\"information\" + 0.005*\"ubuntu\"\n",
      "2017-06-05 20:59:07,303 : INFO : topic #9 (0.067): 0.014*\"money\" + 0.012*\"pay\" + 0.011*\"work\" + 0.009*\"job\" + 0.009*\"jobs\" + 0.007*\"year\" + 0.007*\"company\" + 0.007*\"people\" + 0.006*\"time\" + 0.006*\"id\"\n",
      "2017-06-05 20:59:07,310 : INFO : topic diff=0.112298, rho=0.129099\n",
      "2017-06-05 20:59:07,341 : INFO : PROGRESS: pass 0, at document #122000/246175\n",
      "2017-06-05 20:59:09,009 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:09,343 : INFO : topic #11 (0.067): 0.006*\"energy\" + 0.006*\"city\" + 0.005*\"nuclear\" + 0.005*\"power\" + 0.004*\"solar\" + 0.004*\"global\" + 0.004*\"com jobs\" + 0.003*\"food\" + 0.003*\"water\" + 0.003*\"like\"\n",
      "2017-06-05 20:59:09,345 : INFO : topic #10 (0.067): 0.015*\"code\" + 0.011*\"use\" + 0.008*\"python\" + 0.007*\"onsite\" + 0.005*\"build\" + 0.005*\"using\" + 0.005*\"systems\" + 0.005*\"js\" + 0.005*\"machine\" + 0.005*\"design\"\n",
      "2017-06-05 20:59:09,348 : INFO : topic #8 (0.067): 0.007*\"government\" + 0.007*\"like\" + 0.006*\"right\" + 0.006*\"people\" + 0.006*\"power\" + 0.005*\"san\" + 0.005*\"bad\" + 0.005*\"political\" + 0.005*\"law\" + 0.005*\"francisco\"\n",
      "2017-06-05 20:59:09,350 : INFO : topic #2 (0.067): 0.015*\"software\" + 0.014*\"google\" + 0.012*\"open\" + 0.012*\"product\" + 0.010*\"source\" + 0.009*\"users\" + 0.008*\"make\" + 0.008*\"don\" + 0.008*\"engineer\" + 0.008*\"free\"\n",
      "2017-06-05 20:59:09,353 : INFO : topic #13 (0.067): 0.013*\"problem\" + 0.012*\"time\" + 0.009*\"good\" + 0.008*\"things\" + 0.007*\"just\" + 0.007*\"software\" + 0.006*\"lot\" + 0.006*\"problems\" + 0.006*\"ai\" + 0.006*\"work\"\n",
      "2017-06-05 20:59:09,360 : INFO : topic diff=0.127168, rho=0.128037\n",
      "2017-06-05 20:59:09,395 : INFO : PROGRESS: pass 0, at document #124000/246175\n",
      "2017-06-05 20:59:11,105 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:11,432 : INFO : topic #13 (0.067): 0.012*\"problem\" + 0.012*\"time\" + 0.009*\"good\" + 0.008*\"things\" + 0.007*\"software\" + 0.007*\"just\" + 0.006*\"work\" + 0.006*\"problems\" + 0.006*\"lot\" + 0.006*\"real\"\n",
      "2017-06-05 20:59:11,434 : INFO : topic #12 (0.067): 0.062*\"com\" + 0.039*\"https\" + 0.033*\"www\" + 0.027*\"http\" + 0.020*\"https www\" + 0.012*\"http www\" + 0.010*\"apple\" + 0.010*\"years\" + 0.009*\"ago\" + 0.009*\"like\"\n",
      "2017-06-05 20:59:11,436 : INFO : topic #6 (0.067): 0.017*\"don\" + 0.015*\"just\" + 0.015*\"think\" + 0.013*\"like\" + 0.013*\"work\" + 0.012*\"really\" + 0.012*\"people\" + 0.009*\"need\" + 0.008*\"time\" + 0.007*\"way\"\n",
      "2017-06-05 20:59:11,438 : INFO : topic #0 (0.067): 0.010*\"tesla\" + 0.009*\"car\" + 0.008*\"market\" + 0.008*\"companies\" + 0.007*\"business\" + 0.007*\"people\" + 0.006*\"cars\" + 0.006*\"company\" + 0.006*\"uber\" + 0.005*\"tech\"\n",
      "2017-06-05 20:59:11,440 : INFO : topic #11 (0.067): 0.006*\"energy\" + 0.006*\"city\" + 0.006*\"com jobs\" + 0.005*\"power\" + 0.004*\"solar\" + 0.004*\"global\" + 0.004*\"nuclear\" + 0.004*\"food\" + 0.003*\"york\" + 0.003*\"new york\"\n",
      "2017-06-05 20:59:11,447 : INFO : topic diff=0.105052, rho=0.127000\n",
      "2017-06-05 20:59:11,476 : INFO : PROGRESS: pass 0, at document #126000/246175\n",
      "2017-06-05 20:59:13,102 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:13,425 : INFO : topic #10 (0.067): 0.016*\"code\" + 0.011*\"use\" + 0.008*\"python\" + 0.007*\"onsite\" + 0.005*\"systems\" + 0.005*\"using\" + 0.005*\"build\" + 0.005*\"js\" + 0.005*\"machine\" + 0.005*\"design\"\n",
      "2017-06-05 20:59:13,427 : INFO : topic #13 (0.067): 0.013*\"problem\" + 0.012*\"time\" + 0.009*\"good\" + 0.008*\"things\" + 0.007*\"just\" + 0.006*\"software\" + 0.006*\"work\" + 0.006*\"problems\" + 0.006*\"lot\" + 0.006*\"real\"\n",
      "2017-06-05 20:59:13,429 : INFO : topic #2 (0.067): 0.014*\"software\" + 0.014*\"google\" + 0.012*\"open\" + 0.011*\"product\" + 0.009*\"source\" + 0.008*\"engineer\" + 0.008*\"users\" + 0.008*\"don\" + 0.008*\"make\" + 0.008*\"free\"\n",
      "2017-06-05 20:59:13,431 : INFO : topic #6 (0.067): 0.017*\"don\" + 0.015*\"just\" + 0.015*\"think\" + 0.013*\"like\" + 0.012*\"work\" + 0.012*\"really\" + 0.012*\"people\" + 0.008*\"need\" + 0.008*\"time\" + 0.007*\"way\"\n",
      "2017-06-05 20:59:13,433 : INFO : topic #3 (0.067): 0.011*\"android\" + 0.009*\"use\" + 0.007*\"just\" + 0.006*\"want\" + 0.006*\"desktop\" + 0.006*\"does\" + 0.006*\"phone\" + 0.006*\"ios\" + 0.005*\"user\" + 0.005*\"information\"\n",
      "2017-06-05 20:59:13,440 : INFO : topic diff=0.090951, rho=0.125988\n",
      "2017-06-05 20:59:13,469 : INFO : PROGRESS: pass 0, at document #128000/246175\n",
      "2017-06-05 20:59:15,063 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:15,383 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"people\" + 0.007*\"pro\" + 0.006*\"social\" + 0.006*\"like\" + 0.006*\"post\" + 0.006*\"hn\" + 0.004*\"battery\" + 0.004*\"thing\" + 0.004*\"just\"\n",
      "2017-06-05 20:59:15,385 : INFO : topic #11 (0.067): 0.005*\"city\" + 0.005*\"energy\" + 0.004*\"food\" + 0.004*\"power\" + 0.004*\"com jobs\" + 0.004*\"water\" + 0.003*\"global\" + 0.003*\"article\" + 0.003*\"solar\" + 0.003*\"like\"\n",
      "2017-06-05 20:59:15,388 : INFO : topic #1 (0.067): 0.025*\"people\" + 0.012*\"don\" + 0.008*\"think\" + 0.007*\"doesn\" + 0.007*\"like\" + 0.007*\"know\" + 0.006*\"say\" + 0.006*\"just\" + 0.006*\"point\" + 0.006*\"mean\"\n",
      "2017-06-05 20:59:15,390 : INFO : topic #12 (0.067): 0.061*\"com\" + 0.040*\"https\" + 0.035*\"www\" + 0.027*\"http\" + 0.020*\"https www\" + 0.013*\"http www\" + 0.010*\"apple\" + 0.010*\"years\" + 0.010*\"ago\" + 0.009*\"like\"\n",
      "2017-06-05 20:59:15,392 : INFO : topic #13 (0.067): 0.013*\"problem\" + 0.011*\"time\" + 0.009*\"good\" + 0.008*\"things\" + 0.007*\"just\" + 0.006*\"lot\" + 0.006*\"software\" + 0.006*\"problems\" + 0.006*\"real\" + 0.006*\"work\"\n",
      "2017-06-05 20:59:15,398 : INFO : topic diff=0.100752, rho=0.125000\n",
      "2017-06-05 20:59:15,424 : INFO : PROGRESS: pass 0, at document #130000/246175\n",
      "2017-06-05 20:59:16,987 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:17,304 : INFO : topic #1 (0.067): 0.025*\"people\" + 0.012*\"don\" + 0.008*\"think\" + 0.007*\"doesn\" + 0.007*\"like\" + 0.007*\"know\" + 0.007*\"say\" + 0.006*\"just\" + 0.006*\"point\" + 0.006*\"mean\"\n",
      "2017-06-05 20:59:17,307 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"people\" + 0.007*\"pro\" + 0.006*\"post\" + 0.006*\"like\" + 0.006*\"hn\" + 0.006*\"social\" + 0.004*\"thing\" + 0.004*\"battery\" + 0.004*\"just\"\n",
      "2017-06-05 20:59:17,309 : INFO : topic #3 (0.067): 0.011*\"android\" + 0.009*\"use\" + 0.007*\"just\" + 0.006*\"phone\" + 0.006*\"want\" + 0.006*\"user\" + 0.006*\"does\" + 0.005*\"information\" + 0.005*\"ios\" + 0.005*\"desktop\"\n",
      "2017-06-05 20:59:17,312 : INFO : topic #9 (0.067): 0.013*\"money\" + 0.012*\"jobs\" + 0.011*\"work\" + 0.011*\"pay\" + 0.009*\"job\" + 0.007*\"year\" + 0.007*\"people\" + 0.006*\"news\" + 0.006*\"id\" + 0.006*\"company\"\n",
      "2017-06-05 20:59:17,314 : INFO : topic #5 (0.067): 0.026*\"https\" + 0.015*\"org\" + 0.012*\"com\" + 0.011*\"github\" + 0.010*\"web\" + 0.010*\"windows\" + 0.010*\"en\" + 0.010*\"wikipedia\" + 0.009*\"wiki\" + 0.009*\"use\"\n",
      "2017-06-05 20:59:17,323 : INFO : topic diff=0.084965, rho=0.124035\n",
      "2017-06-05 20:59:17,966 : INFO : PROGRESS: pass 0, at document #132000/246175\n",
      "2017-06-05 20:59:19,527 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:19,835 : INFO : topic #2 (0.067): 0.015*\"google\" + 0.013*\"software\" + 0.013*\"open\" + 0.010*\"product\" + 0.009*\"source\" + 0.008*\"don\" + 0.008*\"free\" + 0.008*\"make\" + 0.008*\"users\" + 0.007*\"engineer\"\n",
      "2017-06-05 20:59:19,837 : INFO : topic #12 (0.067): 0.060*\"com\" + 0.041*\"https\" + 0.036*\"www\" + 0.028*\"http\" + 0.021*\"https www\" + 0.013*\"http www\" + 0.010*\"years\" + 0.010*\"ago\" + 0.009*\"like\" + 0.008*\"apple\"\n",
      "2017-06-05 20:59:19,839 : INFO : topic #6 (0.067): 0.017*\"don\" + 0.016*\"just\" + 0.015*\"think\" + 0.013*\"like\" + 0.012*\"really\" + 0.012*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.008*\"time\" + 0.007*\"way\"\n",
      "2017-06-05 20:59:19,841 : INFO : topic #5 (0.067): 0.027*\"https\" + 0.016*\"org\" + 0.012*\"com\" + 0.011*\"github\" + 0.011*\"en\" + 0.010*\"wikipedia\" + 0.010*\"windows\" + 0.010*\"wiki\" + 0.010*\"web\" + 0.009*\"org wiki\"\n",
      "2017-06-05 20:59:19,844 : INFO : topic #3 (0.067): 0.010*\"android\" + 0.010*\"use\" + 0.007*\"just\" + 0.006*\"phone\" + 0.006*\"want\" + 0.006*\"does\" + 0.006*\"user\" + 0.005*\"desktop\" + 0.005*\"information\" + 0.005*\"ios\"\n",
      "2017-06-05 20:59:19,850 : INFO : topic diff=0.078884, rho=0.123091\n",
      "2017-06-05 20:59:19,877 : INFO : PROGRESS: pass 0, at document #134000/246175\n",
      "2017-06-05 20:59:21,429 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:21,726 : INFO : topic #13 (0.067): 0.013*\"problem\" + 0.011*\"time\" + 0.009*\"good\" + 0.008*\"things\" + 0.007*\"just\" + 0.007*\"ai\" + 0.006*\"lot\" + 0.006*\"software\" + 0.006*\"like\" + 0.006*\"problems\"\n",
      "2017-06-05 20:59:21,728 : INFO : topic #11 (0.067): 0.005*\"energy\" + 0.004*\"city\" + 0.004*\"food\" + 0.004*\"power\" + 0.004*\"water\" + 0.003*\"nuclear\" + 0.003*\"like\" + 0.003*\"article\" + 0.003*\"air\" + 0.003*\"eat\"\n",
      "2017-06-05 20:59:21,731 : INFO : topic #3 (0.067): 0.010*\"use\" + 0.009*\"android\" + 0.007*\"just\" + 0.006*\"desktop\" + 0.006*\"want\" + 0.006*\"phone\" + 0.006*\"user\" + 0.006*\"does\" + 0.005*\"information\" + 0.005*\"ubuntu\"\n",
      "2017-06-05 20:59:21,733 : INFO : topic #5 (0.067): 0.026*\"https\" + 0.016*\"org\" + 0.012*\"com\" + 0.012*\"github\" + 0.010*\"en\" + 0.010*\"windows\" + 0.009*\"wiki\" + 0.009*\"wikipedia\" + 0.009*\"web\" + 0.009*\"use\"\n",
      "2017-06-05 20:59:21,735 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.008*\"people\" + 0.007*\"post\" + 0.006*\"hn\" + 0.006*\"like\" + 0.006*\"pro\" + 0.005*\"social\" + 0.005*\"cancer\" + 0.004*\"thing\" + 0.004*\"just\"\n",
      "2017-06-05 20:59:21,741 : INFO : topic diff=0.077653, rho=0.122169\n",
      "2017-06-05 20:59:21,767 : INFO : PROGRESS: pass 0, at document #136000/246175\n",
      "2017-06-05 20:59:23,351 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:23,656 : INFO : topic #2 (0.067): 0.015*\"google\" + 0.015*\"open\" + 0.013*\"software\" + 0.010*\"source\" + 0.009*\"product\" + 0.008*\"make\" + 0.008*\"don\" + 0.008*\"users\" + 0.008*\"free\" + 0.007*\"open source\"\n",
      "2017-06-05 20:59:23,658 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.008*\"think\" + 0.007*\"doesn\" + 0.007*\"like\" + 0.007*\"say\" + 0.007*\"just\" + 0.007*\"know\" + 0.006*\"point\" + 0.006*\"mean\"\n",
      "2017-06-05 20:59:23,660 : INFO : topic #12 (0.067): 0.060*\"com\" + 0.041*\"https\" + 0.036*\"www\" + 0.028*\"http\" + 0.021*\"https www\" + 0.014*\"http www\" + 0.010*\"ago\" + 0.010*\"years\" + 0.008*\"like\" + 0.007*\"apple\"\n",
      "2017-06-05 20:59:23,663 : INFO : topic #13 (0.067): 0.013*\"problem\" + 0.011*\"time\" + 0.009*\"good\" + 0.008*\"things\" + 0.007*\"just\" + 0.006*\"ai\" + 0.006*\"lot\" + 0.006*\"like\" + 0.006*\"real\" + 0.006*\"software\"\n",
      "2017-06-05 20:59:23,665 : INFO : topic #14 (0.067): 0.026*\"data\" + 0.017*\"like\" + 0.008*\"use\" + 0.006*\"used\" + 0.006*\"type\" + 0.005*\"looks\" + 0.005*\"case\" + 0.005*\"cases\" + 0.005*\"read\" + 0.004*\"change\"\n",
      "2017-06-05 20:59:23,672 : INFO : topic diff=0.076399, rho=0.121268\n",
      "2017-06-05 20:59:23,699 : INFO : PROGRESS: pass 0, at document #138000/246175\n",
      "2017-06-05 20:59:25,340 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:25,637 : INFO : topic #13 (0.067): 0.013*\"problem\" + 0.011*\"time\" + 0.009*\"good\" + 0.008*\"things\" + 0.007*\"just\" + 0.006*\"lot\" + 0.006*\"like\" + 0.006*\"ai\" + 0.006*\"real\" + 0.006*\"software\"\n",
      "2017-06-05 20:59:25,640 : INFO : topic #11 (0.067): 0.005*\"energy\" + 0.004*\"city\" + 0.004*\"military\" + 0.004*\"power\" + 0.004*\"food\" + 0.004*\"wikileaks\" + 0.003*\"water\" + 0.003*\"space\" + 0.003*\"like\" + 0.003*\"nuclear\"\n",
      "2017-06-05 20:59:25,642 : INFO : topic #10 (0.067): 0.018*\"code\" + 0.012*\"use\" + 0.007*\"python\" + 0.006*\"using\" + 0.005*\"systems\" + 0.005*\"language\" + 0.005*\"build\" + 0.004*\"memory\" + 0.004*\"write\" + 0.004*\"time\"\n",
      "2017-06-05 20:59:25,644 : INFO : topic #12 (0.067): 0.059*\"com\" + 0.041*\"https\" + 0.036*\"www\" + 0.028*\"http\" + 0.021*\"https www\" + 0.014*\"http www\" + 0.010*\"ago\" + 0.010*\"years\" + 0.008*\"like\" + 0.007*\"apple\"\n",
      "2017-06-05 20:59:25,647 : INFO : topic #14 (0.067): 0.026*\"data\" + 0.017*\"like\" + 0.008*\"use\" + 0.006*\"used\" + 0.005*\"looks\" + 0.005*\"case\" + 0.005*\"type\" + 0.005*\"cases\" + 0.005*\"read\" + 0.004*\"example\"\n",
      "2017-06-05 20:59:25,655 : INFO : topic diff=0.078169, rho=0.120386\n",
      "2017-06-05 20:59:30,251 : INFO : -9.462 per-word bound, 705.2 perplexity estimate based on a held-out corpus of 2000 documents with 74567 words\n",
      "2017-06-05 20:59:30,252 : INFO : PROGRESS: pass 0, at document #140000/246175\n",
      "2017-06-05 20:59:31,875 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:32,168 : INFO : topic #6 (0.067): 0.017*\"don\" + 0.017*\"just\" + 0.016*\"think\" + 0.014*\"like\" + 0.012*\"really\" + 0.012*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.008*\"way\" + 0.008*\"time\"\n",
      "2017-06-05 20:59:32,171 : INFO : topic #4 (0.067): 0.009*\"life\" + 0.008*\"post\" + 0.008*\"people\" + 0.006*\"hn\" + 0.006*\"like\" + 0.006*\"laptop\" + 0.005*\"social\" + 0.005*\"pro\" + 0.004*\"battery\" + 0.004*\"just\"\n",
      "2017-06-05 20:59:32,173 : INFO : topic #13 (0.067): 0.013*\"problem\" + 0.011*\"time\" + 0.009*\"good\" + 0.008*\"things\" + 0.007*\"just\" + 0.006*\"flavor\" + 0.006*\"lot\" + 0.006*\"like\" + 0.006*\"software\" + 0.006*\"ve\"\n",
      "2017-06-05 20:59:32,175 : INFO : topic #3 (0.067): 0.015*\"desktop\" + 0.013*\"ubuntu\" + 0.010*\"use\" + 0.007*\"android\" + 0.007*\"ubuntu desktop\" + 0.007*\"just\" + 0.007*\"user\" + 0.006*\"want\" + 0.006*\"does\" + 0.006*\"kernel\"\n",
      "2017-06-05 20:59:32,178 : INFO : topic #7 (0.067): 0.011*\"headline\" + 0.007*\"did\" + 0.007*\"war\" + 0.007*\"history\" + 0.005*\"russia\" + 0.004*\"gun\" + 0.004*\"reddit\" + 0.004*\"say\" + 0.004*\"author\" + 0.004*\"wasn\"\n",
      "2017-06-05 20:59:32,184 : INFO : topic diff=0.075002, rho=0.119523\n",
      "2017-06-05 20:59:32,212 : INFO : PROGRESS: pass 0, at document #142000/246175\n",
      "2017-06-05 20:59:33,797 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:34,090 : INFO : topic #2 (0.067): 0.015*\"google\" + 0.014*\"open\" + 0.013*\"software\" + 0.011*\"source\" + 0.008*\"make\" + 0.008*\"product\" + 0.008*\"don\" + 0.008*\"users\" + 0.007*\"internet\" + 0.007*\"use\"\n",
      "2017-06-05 20:59:34,092 : INFO : topic #8 (0.067): 0.007*\"right\" + 0.007*\"government\" + 0.007*\"trump\" + 0.006*\"like\" + 0.006*\"political\" + 0.006*\"people\" + 0.006*\"bad\" + 0.005*\"law\" + 0.005*\"power\" + 0.005*\"state\"\n",
      "2017-06-05 20:59:34,095 : INFO : topic #10 (0.067): 0.018*\"code\" + 0.012*\"use\" + 0.006*\"python\" + 0.006*\"using\" + 0.005*\"systems\" + 0.005*\"build\" + 0.005*\"write\" + 0.005*\"language\" + 0.004*\"memory\" + 0.004*\"design\"\n",
      "2017-06-05 20:59:34,097 : INFO : topic #11 (0.067): 0.005*\"energy\" + 0.004*\"city\" + 0.004*\"space\" + 0.004*\"power\" + 0.004*\"spacex\" + 0.004*\"australia\" + 0.004*\"food\" + 0.003*\"like\" + 0.003*\"military\" + 0.003*\"global\"\n",
      "2017-06-05 20:59:34,099 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.014*\"org\" + 0.011*\"ubuntu\" + 0.011*\"windows\" + 0.010*\"github\" + 0.010*\"com\" + 0.009*\"en\" + 0.009*\"wiki\" + 0.009*\"use\" + 0.009*\"wikipedia\"\n",
      "2017-06-05 20:59:34,106 : INFO : topic diff=0.077045, rho=0.118678\n",
      "2017-06-05 20:59:34,134 : INFO : PROGRESS: pass 0, at document #144000/246175\n",
      "2017-06-05 20:59:35,760 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:36,052 : INFO : topic #2 (0.067): 0.015*\"google\" + 0.013*\"open\" + 0.013*\"software\" + 0.011*\"source\" + 0.008*\"product\" + 0.008*\"make\" + 0.008*\"don\" + 0.008*\"users\" + 0.007*\"use\" + 0.007*\"internet\"\n",
      "2017-06-05 20:59:36,054 : INFO : topic #6 (0.067): 0.017*\"just\" + 0.017*\"don\" + 0.016*\"think\" + 0.014*\"like\" + 0.013*\"really\" + 0.012*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.008*\"time\" + 0.008*\"way\"\n",
      "2017-06-05 20:59:36,057 : INFO : topic #0 (0.067): 0.009*\"business\" + 0.007*\"people\" + 0.007*\"companies\" + 0.007*\"market\" + 0.006*\"cost\" + 0.006*\"tech\" + 0.005*\"company\" + 0.005*\"price\" + 0.005*\"car\" + 0.004*\"just\"\n",
      "2017-06-05 20:59:36,059 : INFO : topic #14 (0.067): 0.024*\"data\" + 0.018*\"like\" + 0.008*\"use\" + 0.006*\"used\" + 0.006*\"description\" + 0.006*\"case\" + 0.006*\"looks\" + 0.005*\"type\" + 0.005*\"cases\" + 0.005*\"read\"\n",
      "2017-06-05 20:59:36,061 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.014*\"org\" + 0.011*\"ubuntu\" + 0.010*\"windows\" + 0.010*\"github\" + 0.010*\"com\" + 0.009*\"en\" + 0.009*\"wikipedia\" + 0.009*\"wiki\" + 0.009*\"use\"\n",
      "2017-06-05 20:59:36,068 : INFO : topic diff=0.071491, rho=0.117851\n",
      "2017-06-05 20:59:36,095 : INFO : PROGRESS: pass 0, at document #146000/246175\n",
      "2017-06-05 20:59:37,709 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:37,992 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.015*\"org\" + 0.010*\"github\" + 0.010*\"windows\" + 0.010*\"com\" + 0.010*\"en\" + 0.010*\"ubuntu\" + 0.009*\"wikipedia\" + 0.009*\"wiki\" + 0.009*\"use\"\n",
      "2017-06-05 20:59:37,994 : INFO : topic #13 (0.067): 0.014*\"problem\" + 0.012*\"time\" + 0.009*\"good\" + 0.009*\"things\" + 0.007*\"just\" + 0.006*\"lot\" + 0.006*\"problems\" + 0.006*\"like\" + 0.006*\"real\" + 0.006*\"ve\"\n",
      "2017-06-05 20:59:37,996 : INFO : topic #3 (0.067): 0.012*\"desktop\" + 0.010*\"use\" + 0.010*\"ubuntu\" + 0.007*\"just\" + 0.006*\"user\" + 0.006*\"want\" + 0.006*\"android\" + 0.006*\"does\" + 0.005*\"ubuntu desktop\" + 0.005*\"information\"\n",
      "2017-06-05 20:59:37,999 : INFO : topic #2 (0.067): 0.015*\"google\" + 0.013*\"open\" + 0.012*\"software\" + 0.011*\"source\" + 0.008*\"product\" + 0.008*\"make\" + 0.008*\"don\" + 0.008*\"users\" + 0.008*\"free\" + 0.007*\"use\"\n",
      "2017-06-05 20:59:38,001 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.008*\"think\" + 0.008*\"doesn\" + 0.007*\"just\" + 0.007*\"say\" + 0.007*\"like\" + 0.007*\"know\" + 0.006*\"point\" + 0.006*\"mean\"\n",
      "2017-06-05 20:59:38,009 : INFO : topic diff=0.070165, rho=0.117041\n",
      "2017-06-05 20:59:38,037 : INFO : PROGRESS: pass 0, at document #148000/246175\n",
      "2017-06-05 20:59:39,621 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:39,903 : INFO : topic #10 (0.067): 0.020*\"code\" + 0.012*\"use\" + 0.006*\"using\" + 0.005*\"language\" + 0.005*\"python\" + 0.005*\"write\" + 0.005*\"systems\" + 0.005*\"build\" + 0.005*\"memory\" + 0.004*\"rust\"\n",
      "2017-06-05 20:59:39,905 : INFO : topic #11 (0.067): 0.007*\"energy\" + 0.006*\"spacex\" + 0.005*\"space\" + 0.005*\"power\" + 0.004*\"drug\" + 0.004*\"rocket\" + 0.004*\"nuclear\" + 0.004*\"water\" + 0.004*\"city\" + 0.003*\"food\"\n",
      "2017-06-05 20:59:39,908 : INFO : topic #9 (0.067): 0.015*\"money\" + 0.012*\"pay\" + 0.011*\"work\" + 0.008*\"year\" + 0.007*\"jobs\" + 0.007*\"job\" + 0.007*\"news\" + 0.007*\"people\" + 0.006*\"tax\" + 0.005*\"time\"\n",
      "2017-06-05 20:59:39,910 : INFO : topic #2 (0.067): 0.015*\"google\" + 0.013*\"open\" + 0.012*\"software\" + 0.011*\"source\" + 0.008*\"free\" + 0.008*\"users\" + 0.008*\"make\" + 0.008*\"don\" + 0.007*\"product\" + 0.007*\"use\"\n",
      "2017-06-05 20:59:39,912 : INFO : topic #3 (0.067): 0.011*\"desktop\" + 0.011*\"use\" + 0.009*\"ubuntu\" + 0.007*\"just\" + 0.006*\"user\" + 0.006*\"want\" + 0.006*\"does\" + 0.006*\"information\" + 0.005*\"android\" + 0.005*\"ubuntu desktop\"\n",
      "2017-06-05 20:59:39,919 : INFO : topic diff=0.066436, rho=0.116248\n",
      "2017-06-05 20:59:39,945 : INFO : PROGRESS: pass 0, at document #150000/246175\n",
      "2017-06-05 20:59:41,510 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:41,785 : INFO : topic #12 (0.067): 0.055*\"com\" + 0.040*\"https\" + 0.036*\"www\" + 0.028*\"http\" + 0.021*\"https www\" + 0.015*\"http www\" + 0.011*\"ago\" + 0.010*\"years\" + 0.008*\"like\" + 0.007*\"watch\"\n",
      "2017-06-05 20:59:41,788 : INFO : topic #0 (0.067): 0.008*\"business\" + 0.008*\"companies\" + 0.007*\"cost\" + 0.007*\"market\" + 0.007*\"people\" + 0.007*\"price\" + 0.005*\"company\" + 0.005*\"car\" + 0.005*\"tech\" + 0.004*\"just\"\n",
      "2017-06-05 20:59:41,790 : INFO : topic #9 (0.067): 0.016*\"money\" + 0.012*\"tax\" + 0.012*\"pay\" + 0.010*\"work\" + 0.008*\"year\" + 0.007*\"jobs\" + 0.007*\"people\" + 0.006*\"job\" + 0.006*\"news\" + 0.006*\"income\"\n",
      "2017-06-05 20:59:41,792 : INFO : topic #13 (0.067): 0.014*\"problem\" + 0.011*\"time\" + 0.009*\"good\" + 0.009*\"things\" + 0.007*\"just\" + 0.006*\"problems\" + 0.006*\"lot\" + 0.006*\"like\" + 0.006*\"ve\" + 0.006*\"better\"\n",
      "2017-06-05 20:59:41,795 : INFO : topic #11 (0.067): 0.007*\"energy\" + 0.005*\"spacex\" + 0.005*\"space\" + 0.005*\"power\" + 0.004*\"nuclear\" + 0.004*\"food\" + 0.004*\"drug\" + 0.004*\"water\" + 0.004*\"rocket\" + 0.004*\"eat\"\n",
      "2017-06-05 20:59:41,801 : INFO : topic diff=0.069184, rho=0.115470\n",
      "2017-06-05 20:59:41,827 : INFO : PROGRESS: pass 0, at document #152000/246175\n",
      "2017-06-05 20:59:43,439 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:43,712 : INFO : topic #7 (0.067): 0.008*\"history\" + 0.007*\"did\" + 0.006*\"war\" + 0.005*\"headline\" + 0.004*\"author\" + 0.004*\"wasn\" + 0.004*\"say\" + 0.004*\"russia\" + 0.004*\"women\" + 0.004*\"read\"\n",
      "2017-06-05 20:59:43,715 : INFO : topic #5 (0.067): 0.025*\"https\" + 0.017*\"org\" + 0.011*\"en\" + 0.010*\"wikipedia\" + 0.010*\"wiki\" + 0.010*\"github\" + 0.010*\"com\" + 0.009*\"windows\" + 0.009*\"org wiki\" + 0.009*\"use\"\n",
      "2017-06-05 20:59:43,717 : INFO : topic #8 (0.067): 0.008*\"government\" + 0.007*\"law\" + 0.006*\"right\" + 0.006*\"power\" + 0.006*\"people\" + 0.006*\"like\" + 0.005*\"political\" + 0.005*\"trump\" + 0.005*\"state\" + 0.005*\"bad\"\n",
      "2017-06-05 20:59:43,720 : INFO : topic #9 (0.067): 0.015*\"money\" + 0.012*\"pay\" + 0.011*\"tax\" + 0.010*\"work\" + 0.008*\"year\" + 0.007*\"people\" + 0.007*\"jobs\" + 0.006*\"job\" + 0.006*\"news\" + 0.006*\"income\"\n",
      "2017-06-05 20:59:43,723 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.007*\"people\" + 0.007*\"hn\" + 0.006*\"like\" + 0.005*\"privacy\" + 0.005*\"social\" + 0.004*\"just\" + 0.004*\"thing\" + 0.004*\"ve\"\n",
      "2017-06-05 20:59:43,729 : INFO : topic diff=0.071567, rho=0.114708\n",
      "2017-06-05 20:59:43,756 : INFO : PROGRESS: pass 0, at document #154000/246175\n",
      "2017-06-05 20:59:45,350 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:45,631 : INFO : topic #5 (0.067): 0.025*\"https\" + 0.017*\"org\" + 0.011*\"en\" + 0.011*\"wikipedia\" + 0.010*\"wiki\" + 0.010*\"github\" + 0.009*\"com\" + 0.009*\"org wiki\" + 0.009*\"https en\" + 0.009*\"use\"\n",
      "2017-06-05 20:59:45,633 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.008*\"doesn\" + 0.007*\"say\" + 0.007*\"just\" + 0.007*\"point\" + 0.006*\"like\" + 0.006*\"mean\" + 0.006*\"know\"\n",
      "2017-06-05 20:59:45,635 : INFO : topic #14 (0.067): 0.027*\"data\" + 0.017*\"like\" + 0.008*\"use\" + 0.006*\"read\" + 0.006*\"looks\" + 0.006*\"used\" + 0.005*\"case\" + 0.005*\"type\" + 0.005*\"cases\" + 0.005*\"change\"\n",
      "2017-06-05 20:59:45,637 : INFO : topic #8 (0.067): 0.008*\"government\" + 0.007*\"law\" + 0.007*\"right\" + 0.006*\"power\" + 0.006*\"people\" + 0.006*\"like\" + 0.005*\"political\" + 0.005*\"state\" + 0.005*\"trump\" + 0.005*\"bad\"\n",
      "2017-06-05 20:59:45,639 : INFO : topic #3 (0.067): 0.011*\"use\" + 0.009*\"desktop\" + 0.007*\"just\" + 0.007*\"vpn\" + 0.006*\"ubuntu\" + 0.006*\"user\" + 0.006*\"information\" + 0.006*\"want\" + 0.006*\"isp\" + 0.006*\"does\"\n",
      "2017-06-05 20:59:45,646 : INFO : topic diff=0.066863, rho=0.113961\n",
      "2017-06-05 20:59:45,672 : INFO : PROGRESS: pass 0, at document #156000/246175\n",
      "2017-06-05 20:59:47,208 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:47,475 : INFO : topic #14 (0.067): 0.027*\"data\" + 0.017*\"like\" + 0.009*\"use\" + 0.007*\"read\" + 0.006*\"looks\" + 0.006*\"case\" + 0.006*\"used\" + 0.006*\"type\" + 0.005*\"cases\" + 0.005*\"change\"\n",
      "2017-06-05 20:59:47,478 : INFO : topic #8 (0.067): 0.009*\"government\" + 0.007*\"law\" + 0.007*\"right\" + 0.006*\"power\" + 0.006*\"people\" + 0.006*\"like\" + 0.005*\"political\" + 0.005*\"trump\" + 0.005*\"state\" + 0.005*\"bad\"\n",
      "2017-06-05 20:59:47,480 : INFO : topic #10 (0.067): 0.019*\"code\" + 0.012*\"use\" + 0.006*\"using\" + 0.005*\"language\" + 0.005*\"build\" + 0.005*\"systems\" + 0.004*\"write\" + 0.004*\"memory\" + 0.004*\"python\" + 0.004*\"machine\"\n",
      "2017-06-05 20:59:47,482 : INFO : topic #0 (0.067): 0.008*\"business\" + 0.008*\"companies\" + 0.008*\"cost\" + 0.007*\"market\" + 0.007*\"people\" + 0.006*\"price\" + 0.006*\"company\" + 0.005*\"car\" + 0.005*\"tech\" + 0.005*\"long\"\n",
      "2017-06-05 20:59:47,484 : INFO : topic #9 (0.067): 0.014*\"money\" + 0.012*\"pay\" + 0.011*\"work\" + 0.009*\"tax\" + 0.009*\"eu\" + 0.008*\"uk\" + 0.008*\"year\" + 0.007*\"people\" + 0.006*\"jobs\" + 0.006*\"job\"\n",
      "2017-06-05 20:59:47,491 : INFO : topic diff=0.065735, rho=0.113228\n",
      "2017-06-05 20:59:47,517 : INFO : PROGRESS: pass 0, at document #158000/246175\n",
      "2017-06-05 20:59:49,087 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:49,354 : INFO : topic #14 (0.067): 0.027*\"data\" + 0.017*\"like\" + 0.008*\"use\" + 0.007*\"read\" + 0.006*\"looks\" + 0.006*\"case\" + 0.006*\"used\" + 0.005*\"cases\" + 0.005*\"type\" + 0.005*\"change\"\n",
      "2017-06-05 20:59:49,356 : INFO : topic #12 (0.067): 0.054*\"com\" + 0.042*\"https\" + 0.037*\"www\" + 0.030*\"http\" + 0.021*\"https www\" + 0.015*\"http www\" + 0.011*\"ago\" + 0.010*\"years\" + 0.008*\"like\" + 0.008*\"news\"\n",
      "2017-06-05 20:59:49,358 : INFO : topic #3 (0.067): 0.011*\"use\" + 0.008*\"vpn\" + 0.008*\"just\" + 0.007*\"desktop\" + 0.007*\"isp\" + 0.007*\"user\" + 0.006*\"information\" + 0.006*\"isps\" + 0.006*\"does\" + 0.006*\"want\"\n",
      "2017-06-05 20:59:49,360 : INFO : topic #2 (0.067): 0.015*\"google\" + 0.013*\"open\" + 0.011*\"source\" + 0.011*\"software\" + 0.008*\"users\" + 0.008*\"free\" + 0.008*\"internet\" + 0.008*\"don\" + 0.008*\"product\" + 0.008*\"make\"\n",
      "2017-06-05 20:59:49,363 : INFO : topic #11 (0.067): 0.008*\"nuclear\" + 0.007*\"energy\" + 0.006*\"power\" + 0.004*\"space\" + 0.004*\"solar\" + 0.004*\"spacex\" + 0.004*\"uk\" + 0.004*\"food\" + 0.003*\"water\" + 0.003*\"world\"\n",
      "2017-06-05 20:59:49,370 : INFO : topic diff=0.057222, rho=0.112509\n",
      "2017-06-05 20:59:53,816 : INFO : -9.458 per-word bound, 703.5 perplexity estimate based on a held-out corpus of 2000 documents with 72470 words\n",
      "2017-06-05 20:59:53,817 : INFO : PROGRESS: pass 0, at document #160000/246175\n",
      "2017-06-05 20:59:55,403 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:55,671 : INFO : topic #0 (0.067): 0.008*\"business\" + 0.008*\"market\" + 0.008*\"companies\" + 0.008*\"cost\" + 0.007*\"people\" + 0.006*\"car\" + 0.006*\"company\" + 0.006*\"price\" + 0.006*\"tech\" + 0.005*\"tesla\"\n",
      "2017-06-05 20:59:55,673 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.008*\"doesn\" + 0.007*\"say\" + 0.007*\"just\" + 0.007*\"point\" + 0.006*\"mean\" + 0.006*\"like\" + 0.006*\"know\"\n",
      "2017-06-05 20:59:55,676 : INFO : topic #6 (0.067): 0.019*\"don\" + 0.018*\"just\" + 0.017*\"think\" + 0.014*\"like\" + 0.013*\"really\" + 0.012*\"people\" + 0.012*\"work\" + 0.009*\"need\" + 0.008*\"time\" + 0.008*\"way\"\n",
      "2017-06-05 20:59:55,678 : INFO : topic #14 (0.067): 0.026*\"data\" + 0.017*\"like\" + 0.008*\"use\" + 0.007*\"read\" + 0.006*\"type\" + 0.006*\"used\" + 0.006*\"looks\" + 0.006*\"case\" + 0.005*\"cases\" + 0.005*\"change\"\n",
      "2017-06-05 20:59:55,680 : INFO : topic #9 (0.067): 0.013*\"money\" + 0.012*\"pay\" + 0.011*\"work\" + 0.008*\"tax\" + 0.008*\"year\" + 0.008*\"eu\" + 0.008*\"jobs\" + 0.007*\"uk\" + 0.007*\"people\" + 0.007*\"job\"\n",
      "2017-06-05 20:59:55,686 : INFO : topic diff=0.061166, rho=0.111803\n",
      "2017-06-05 20:59:55,712 : INFO : PROGRESS: pass 0, at document #162000/246175\n",
      "2017-06-05 20:59:57,290 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:57,550 : INFO : topic #6 (0.067): 0.019*\"don\" + 0.018*\"just\" + 0.017*\"think\" + 0.015*\"like\" + 0.013*\"really\" + 0.012*\"people\" + 0.012*\"work\" + 0.009*\"need\" + 0.008*\"time\" + 0.008*\"way\"\n",
      "2017-06-05 20:59:57,553 : INFO : topic #5 (0.067): 0.025*\"https\" + 0.016*\"org\" + 0.012*\"wikipedia\" + 0.011*\"en\" + 0.011*\"github\" + 0.011*\"wiki\" + 0.010*\"com\" + 0.009*\"org wiki\" + 0.009*\"https en\" + 0.009*\"wikipedia org\"\n",
      "2017-06-05 20:59:57,555 : INFO : topic #8 (0.067): 0.008*\"government\" + 0.007*\"law\" + 0.006*\"state\" + 0.006*\"right\" + 0.006*\"people\" + 0.006*\"power\" + 0.006*\"trump\" + 0.006*\"like\" + 0.005*\"political\" + 0.005*\"vote\"\n",
      "2017-06-05 20:59:57,558 : INFO : topic #2 (0.067): 0.014*\"google\" + 0.013*\"open\" + 0.011*\"source\" + 0.010*\"software\" + 0.009*\"free\" + 0.008*\"users\" + 0.008*\"don\" + 0.008*\"use\" + 0.008*\"product\" + 0.008*\"make\"\n",
      "2017-06-05 20:59:57,559 : INFO : topic #14 (0.067): 0.026*\"data\" + 0.017*\"like\" + 0.008*\"use\" + 0.007*\"read\" + 0.006*\"type\" + 0.006*\"looks\" + 0.006*\"used\" + 0.006*\"case\" + 0.005*\"change\" + 0.005*\"cases\"\n",
      "2017-06-05 20:59:57,566 : INFO : topic diff=0.054514, rho=0.111111\n",
      "2017-06-05 20:59:57,591 : INFO : PROGRESS: pass 0, at document #164000/246175\n",
      "2017-06-05 20:59:59,112 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 20:59:59,366 : INFO : topic #5 (0.067): 0.025*\"https\" + 0.017*\"org\" + 0.012*\"wikipedia\" + 0.011*\"en\" + 0.011*\"github\" + 0.011*\"wiki\" + 0.010*\"com\" + 0.010*\"org wiki\" + 0.009*\"wikipedia org\" + 0.009*\"https en\"\n",
      "2017-06-05 20:59:59,369 : INFO : topic #10 (0.067): 0.020*\"code\" + 0.012*\"use\" + 0.006*\"using\" + 0.006*\"language\" + 0.005*\"build\" + 0.005*\"python\" + 0.005*\"write\" + 0.005*\"memory\" + 0.004*\"machine\" + 0.004*\"systems\"\n",
      "2017-06-05 20:59:59,370 : INFO : topic #2 (0.067): 0.014*\"google\" + 0.012*\"open\" + 0.011*\"source\" + 0.010*\"software\" + 0.008*\"free\" + 0.008*\"users\" + 0.008*\"product\" + 0.008*\"don\" + 0.008*\"use\" + 0.008*\"make\"\n",
      "2017-06-05 20:59:59,373 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.007*\"people\" + 0.007*\"hn\" + 0.007*\"post\" + 0.006*\"privacy\" + 0.006*\"like\" + 0.005*\"social\" + 0.005*\"friends\" + 0.004*\"ve\" + 0.004*\"just\"\n",
      "2017-06-05 20:59:59,374 : INFO : topic #12 (0.067): 0.055*\"com\" + 0.040*\"https\" + 0.037*\"www\" + 0.029*\"http\" + 0.020*\"https www\" + 0.016*\"http www\" + 0.012*\"ago\" + 0.010*\"years\" + 0.008*\"like\" + 0.008*\"news\"\n",
      "2017-06-05 20:59:59,381 : INFO : topic diff=0.054519, rho=0.110432\n",
      "2017-06-05 20:59:59,407 : INFO : PROGRESS: pass 0, at document #166000/246175\n",
      "2017-06-05 21:00:00,949 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:01,203 : INFO : topic #6 (0.067): 0.020*\"don\" + 0.018*\"just\" + 0.017*\"think\" + 0.015*\"like\" + 0.013*\"really\" + 0.012*\"people\" + 0.012*\"work\" + 0.009*\"need\" + 0.008*\"way\" + 0.008*\"time\"\n",
      "2017-06-05 21:00:01,206 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.008*\"doesn\" + 0.007*\"say\" + 0.007*\"just\" + 0.007*\"point\" + 0.006*\"know\" + 0.006*\"like\" + 0.006*\"mean\"\n",
      "2017-06-05 21:00:01,208 : INFO : topic #3 (0.067): 0.011*\"use\" + 0.008*\"just\" + 0.007*\"vpn\" + 0.007*\"user\" + 0.006*\"does\" + 0.006*\"information\" + 0.006*\"want\" + 0.005*\"isp\" + 0.005*\"desktop\" + 0.004*\"isps\"\n",
      "2017-06-05 21:00:01,211 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.007*\"hn\" + 0.007*\"people\" + 0.007*\"post\" + 0.006*\"like\" + 0.006*\"privacy\" + 0.005*\"friends\" + 0.005*\"social\" + 0.004*\"just\" + 0.004*\"ve\"\n",
      "2017-06-05 21:00:01,213 : INFO : topic #10 (0.067): 0.020*\"code\" + 0.012*\"use\" + 0.007*\"language\" + 0.006*\"using\" + 0.005*\"memory\" + 0.005*\"build\" + 0.005*\"python\" + 0.005*\"write\" + 0.004*\"machine\" + 0.004*\"systems\"\n",
      "2017-06-05 21:00:01,220 : INFO : topic diff=0.054616, rho=0.109764\n",
      "2017-06-05 21:00:01,248 : INFO : PROGRESS: pass 0, at document #168000/246175\n",
      "2017-06-05 21:00:02,825 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:03,082 : INFO : topic #9 (0.067): 0.013*\"money\" + 0.013*\"pay\" + 0.011*\"work\" + 0.008*\"year\" + 0.007*\"people\" + 0.007*\"job\" + 0.007*\"jobs\" + 0.007*\"tax\" + 0.006*\"uk\" + 0.006*\"news\"\n",
      "2017-06-05 21:00:03,085 : INFO : topic #8 (0.067): 0.008*\"government\" + 0.006*\"law\" + 0.006*\"right\" + 0.006*\"state\" + 0.006*\"people\" + 0.006*\"power\" + 0.005*\"like\" + 0.005*\"political\" + 0.005*\"bad\" + 0.005*\"trump\"\n",
      "2017-06-05 21:00:03,087 : INFO : topic #7 (0.067): 0.012*\"women\" + 0.008*\"war\" + 0.007*\"history\" + 0.007*\"did\" + 0.004*\"male\" + 0.004*\"read\" + 0.004*\"say\" + 0.004*\"wasn\" + 0.004*\"reddit\" + 0.004*\"female\"\n",
      "2017-06-05 21:00:03,089 : INFO : topic #11 (0.067): 0.007*\"drug\" + 0.006*\"energy\" + 0.006*\"nuclear\" + 0.005*\"power\" + 0.004*\"space\" + 0.004*\"water\" + 0.004*\"food\" + 0.004*\"solar\" + 0.003*\"world\" + 0.003*\"new\"\n",
      "2017-06-05 21:00:03,092 : INFO : topic #5 (0.067): 0.025*\"https\" + 0.016*\"org\" + 0.012*\"wikipedia\" + 0.011*\"en\" + 0.011*\"github\" + 0.010*\"wiki\" + 0.010*\"com\" + 0.009*\"org wiki\" + 0.009*\"windows\" + 0.009*\"https en\"\n",
      "2017-06-05 21:00:03,098 : INFO : topic diff=0.061646, rho=0.109109\n",
      "2017-06-05 21:00:03,126 : INFO : PROGRESS: pass 0, at document #170000/246175\n",
      "2017-06-05 21:00:04,841 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:05,098 : INFO : topic #2 (0.067): 0.014*\"google\" + 0.012*\"open\" + 0.010*\"content\" + 0.010*\"facebook\" + 0.010*\"software\" + 0.009*\"source\" + 0.008*\"free\" + 0.008*\"don\" + 0.008*\"users\" + 0.008*\"make\"\n",
      "2017-06-05 21:00:05,100 : INFO : topic #14 (0.067): 0.023*\"data\" + 0.017*\"like\" + 0.008*\"use\" + 0.007*\"read\" + 0.007*\"type\" + 0.006*\"looks\" + 0.006*\"used\" + 0.006*\"case\" + 0.005*\"change\" + 0.005*\"cases\"\n",
      "2017-06-05 21:00:05,102 : INFO : topic #11 (0.067): 0.006*\"drug\" + 0.006*\"energy\" + 0.005*\"power\" + 0.005*\"nuclear\" + 0.004*\"water\" + 0.004*\"space\" + 0.004*\"food\" + 0.004*\"world\" + 0.003*\"solar\" + 0.003*\"new\"\n",
      "2017-06-05 21:00:05,104 : INFO : topic #6 (0.067): 0.020*\"don\" + 0.018*\"just\" + 0.017*\"think\" + 0.015*\"like\" + 0.013*\"really\" + 0.012*\"people\" + 0.011*\"work\" + 0.008*\"need\" + 0.008*\"way\" + 0.008*\"time\"\n",
      "2017-06-05 21:00:05,106 : INFO : topic #13 (0.067): 0.014*\"problem\" + 0.012*\"time\" + 0.010*\"good\" + 0.010*\"things\" + 0.008*\"just\" + 0.007*\"lot\" + 0.007*\"problems\" + 0.007*\"ve\" + 0.006*\"like\" + 0.006*\"better\"\n",
      "2017-06-05 21:00:05,113 : INFO : topic diff=0.052628, rho=0.108465\n",
      "2017-06-05 21:00:05,139 : INFO : PROGRESS: pass 0, at document #172000/246175\n",
      "2017-06-05 21:00:06,702 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:06,968 : INFO : topic #13 (0.067): 0.015*\"problem\" + 0.012*\"time\" + 0.010*\"good\" + 0.010*\"things\" + 0.007*\"just\" + 0.007*\"lot\" + 0.007*\"ve\" + 0.007*\"problems\" + 0.006*\"game\" + 0.006*\"like\"\n",
      "2017-06-05 21:00:06,970 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.007*\"people\" + 0.007*\"hn\" + 0.006*\"like\" + 0.006*\"social\" + 0.005*\"privacy\" + 0.005*\"friends\" + 0.005*\"ve\" + 0.004*\"laptop\"\n",
      "2017-06-05 21:00:06,973 : INFO : topic #5 (0.067): 0.025*\"https\" + 0.016*\"org\" + 0.012*\"wikipedia\" + 0.011*\"en\" + 0.010*\"github\" + 0.010*\"wiki\" + 0.010*\"com\" + 0.009*\"org wiki\" + 0.009*\"https en\" + 0.009*\"wikipedia org\"\n",
      "2017-06-05 21:00:06,976 : INFO : topic #7 (0.067): 0.010*\"women\" + 0.007*\"war\" + 0.007*\"did\" + 0.007*\"coins\" + 0.007*\"history\" + 0.004*\"wasn\" + 0.004*\"chinese\" + 0.004*\"read\" + 0.004*\"reddit\" + 0.004*\"say\"\n",
      "2017-06-05 21:00:06,978 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.008*\"doesn\" + 0.007*\"say\" + 0.007*\"point\" + 0.007*\"just\" + 0.006*\"like\" + 0.006*\"know\" + 0.006*\"mean\"\n",
      "2017-06-05 21:00:06,985 : INFO : topic diff=0.052265, rho=0.107833\n",
      "2017-06-05 21:00:07,010 : INFO : PROGRESS: pass 0, at document #174000/246175\n",
      "2017-06-05 21:00:08,610 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:08,860 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.008*\"post\" + 0.007*\"people\" + 0.007*\"hn\" + 0.006*\"like\" + 0.006*\"social\" + 0.005*\"privacy\" + 0.005*\"search\" + 0.005*\"ve\" + 0.005*\"friends\"\n",
      "2017-06-05 21:00:08,862 : INFO : topic #10 (0.067): 0.020*\"code\" + 0.012*\"use\" + 0.008*\"language\" + 0.006*\"using\" + 0.005*\"rust\" + 0.005*\"write\" + 0.005*\"memory\" + 0.005*\"python\" + 0.004*\"build\" + 0.004*\"js\"\n",
      "2017-06-05 21:00:08,865 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.008*\"doesn\" + 0.007*\"say\" + 0.007*\"point\" + 0.007*\"just\" + 0.006*\"like\" + 0.006*\"know\" + 0.006*\"mean\"\n",
      "2017-06-05 21:00:08,867 : INFO : topic #3 (0.067): 0.011*\"use\" + 0.008*\"just\" + 0.007*\"user\" + 0.006*\"does\" + 0.006*\"information\" + 0.006*\"want\" + 0.006*\"desktop\" + 0.005*\"account\" + 0.005*\"vpn\" + 0.004*\"traffic\"\n",
      "2017-06-05 21:00:08,869 : INFO : topic #8 (0.067): 0.008*\"government\" + 0.006*\"law\" + 0.006*\"state\" + 0.006*\"right\" + 0.006*\"power\" + 0.006*\"people\" + 0.006*\"bad\" + 0.005*\"like\" + 0.005*\"political\" + 0.005*\"states\"\n",
      "2017-06-05 21:00:08,875 : INFO : topic diff=0.048228, rho=0.107211\n",
      "2017-06-05 21:00:08,902 : INFO : PROGRESS: pass 0, at document #176000/246175\n",
      "2017-06-05 21:00:10,450 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:10,691 : INFO : topic #11 (0.067): 0.006*\"energy\" + 0.005*\"drug\" + 0.005*\"power\" + 0.005*\"nuclear\" + 0.004*\"space\" + 0.004*\"world\" + 0.004*\"water\" + 0.004*\"food\" + 0.003*\"global\" + 0.003*\"climate\"\n",
      "2017-06-05 21:00:10,694 : INFO : topic #13 (0.067): 0.014*\"problem\" + 0.012*\"time\" + 0.010*\"good\" + 0.010*\"things\" + 0.007*\"just\" + 0.007*\"ve\" + 0.007*\"lot\" + 0.007*\"problems\" + 0.006*\"game\" + 0.006*\"better\"\n",
      "2017-06-05 21:00:10,697 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.008*\"post\" + 0.007*\"people\" + 0.007*\"hn\" + 0.006*\"like\" + 0.005*\"social\" + 0.005*\"search\" + 0.005*\"privacy\" + 0.005*\"ve\" + 0.005*\"friends\"\n",
      "2017-06-05 21:00:10,699 : INFO : topic #9 (0.067): 0.013*\"pay\" + 0.012*\"money\" + 0.011*\"work\" + 0.009*\"job\" + 0.008*\"year\" + 0.007*\"people\" + 0.007*\"jobs\" + 0.006*\"uk\" + 0.006*\"tax\" + 0.006*\"news\"\n",
      "2017-06-05 21:00:10,701 : INFO : topic #12 (0.067): 0.054*\"com\" + 0.041*\"https\" + 0.037*\"www\" + 0.029*\"http\" + 0.022*\"https www\" + 0.015*\"http www\" + 0.011*\"ago\" + 0.010*\"youtube\" + 0.009*\"years\" + 0.009*\"watch\"\n",
      "2017-06-05 21:00:10,707 : INFO : topic diff=0.051010, rho=0.106600\n",
      "2017-06-05 21:00:10,734 : INFO : PROGRESS: pass 0, at document #178000/246175\n",
      "2017-06-05 21:00:12,280 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:12,517 : INFO : topic #14 (0.067): 0.023*\"data\" + 0.017*\"like\" + 0.009*\"use\" + 0.008*\"type\" + 0.007*\"read\" + 0.006*\"used\" + 0.006*\"looks\" + 0.006*\"function\" + 0.005*\"case\" + 0.005*\"example\"\n",
      "2017-06-05 21:00:12,519 : INFO : topic #12 (0.067): 0.054*\"com\" + 0.040*\"https\" + 0.036*\"www\" + 0.029*\"http\" + 0.021*\"https www\" + 0.015*\"http www\" + 0.012*\"ago\" + 0.010*\"amazon\" + 0.010*\"years\" + 0.009*\"youtube\"\n",
      "2017-06-05 21:00:12,521 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.008*\"doesn\" + 0.008*\"say\" + 0.007*\"point\" + 0.007*\"just\" + 0.006*\"know\" + 0.006*\"like\" + 0.006*\"mean\"\n",
      "2017-06-05 21:00:12,523 : INFO : topic #9 (0.067): 0.012*\"pay\" + 0.011*\"work\" + 0.011*\"money\" + 0.010*\"job\" + 0.009*\"year\" + 0.008*\"jobs\" + 0.007*\"people\" + 0.005*\"item\" + 0.005*\"news\" + 0.005*\"tax\"\n",
      "2017-06-05 21:00:12,526 : INFO : topic #5 (0.067): 0.024*\"https\" + 0.016*\"org\" + 0.011*\"wikipedia\" + 0.011*\"en\" + 0.010*\"github\" + 0.010*\"wiki\" + 0.009*\"com\" + 0.009*\"org wiki\" + 0.009*\"https en\" + 0.009*\"web\"\n",
      "2017-06-05 21:00:12,532 : INFO : topic diff=0.047973, rho=0.106000\n",
      "2017-06-05 21:00:16,958 : INFO : -9.516 per-word bound, 732.4 perplexity estimate based on a held-out corpus of 2000 documents with 69276 words\n",
      "2017-06-05 21:00:16,959 : INFO : PROGRESS: pass 0, at document #180000/246175\n",
      "2017-06-05 21:00:18,539 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:18,785 : INFO : topic #5 (0.067): 0.024*\"https\" + 0.016*\"org\" + 0.012*\"wikipedia\" + 0.011*\"en\" + 0.010*\"wiki\" + 0.010*\"github\" + 0.009*\"org wiki\" + 0.009*\"com\" + 0.009*\"windows\" + 0.009*\"https en\"\n",
      "2017-06-05 21:00:18,787 : INFO : topic #12 (0.067): 0.053*\"com\" + 0.040*\"https\" + 0.036*\"www\" + 0.029*\"http\" + 0.021*\"https www\" + 0.015*\"http www\" + 0.012*\"ago\" + 0.010*\"amazon\" + 0.010*\"years\" + 0.008*\"youtube\"\n",
      "2017-06-05 21:00:18,789 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.007*\"post\" + 0.007*\"people\" + 0.007*\"hn\" + 0.006*\"like\" + 0.005*\"social\" + 0.005*\"battery\" + 0.005*\"ve\" + 0.005*\"laptop\" + 0.005*\"search\"\n",
      "2017-06-05 21:00:18,791 : INFO : topic #9 (0.067): 0.012*\"pay\" + 0.012*\"work\" + 0.011*\"money\" + 0.010*\"job\" + 0.009*\"year\" + 0.008*\"jobs\" + 0.007*\"people\" + 0.006*\"tax\" + 0.006*\"item\" + 0.006*\"news\"\n",
      "2017-06-05 21:00:18,794 : INFO : topic #6 (0.067): 0.020*\"don\" + 0.018*\"just\" + 0.018*\"think\" + 0.015*\"like\" + 0.013*\"really\" + 0.013*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.008*\"know\" + 0.008*\"way\"\n",
      "2017-06-05 21:00:18,801 : INFO : topic diff=0.045592, rho=0.105409\n",
      "2017-06-05 21:00:18,829 : INFO : PROGRESS: pass 0, at document #182000/246175\n",
      "2017-06-05 21:00:20,530 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:20,768 : INFO : topic #3 (0.067): 0.011*\"use\" + 0.008*\"just\" + 0.007*\"user\" + 0.006*\"does\" + 0.006*\"want\" + 0.006*\"phone\" + 0.006*\"desktop\" + 0.005*\"account\" + 0.005*\"information\" + 0.004*\"don\"\n",
      "2017-06-05 21:00:20,770 : INFO : topic #7 (0.067): 0.009*\"women\" + 0.008*\"did\" + 0.007*\"war\" + 0.007*\"history\" + 0.005*\"chinese\" + 0.005*\"china\" + 0.004*\"coins\" + 0.004*\"author\" + 0.004*\"say\" + 0.004*\"read\"\n",
      "2017-06-05 21:00:20,772 : INFO : topic #12 (0.067): 0.054*\"com\" + 0.040*\"https\" + 0.036*\"www\" + 0.029*\"http\" + 0.021*\"https www\" + 0.015*\"http www\" + 0.012*\"ago\" + 0.010*\"amazon\" + 0.010*\"years\" + 0.008*\"youtube\"\n",
      "2017-06-05 21:00:20,775 : INFO : topic #11 (0.067): 0.005*\"energy\" + 0.005*\"power\" + 0.005*\"drug\" + 0.004*\"space\" + 0.004*\"air\" + 0.004*\"world\" + 0.004*\"nuclear\" + 0.004*\"global\" + 0.003*\"water\" + 0.003*\"food\"\n",
      "2017-06-05 21:00:20,777 : INFO : topic #2 (0.067): 0.021*\"google\" + 0.012*\"open\" + 0.009*\"product\" + 0.009*\"software\" + 0.008*\"don\" + 0.008*\"content\" + 0.008*\"users\" + 0.008*\"use\" + 0.008*\"facebook\" + 0.008*\"source\"\n",
      "2017-06-05 21:00:20,784 : INFO : topic diff=0.050037, rho=0.104828\n",
      "2017-06-05 21:00:20,808 : INFO : PROGRESS: pass 0, at document #184000/246175\n",
      "2017-06-05 21:00:22,385 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:22,618 : INFO : topic #9 (0.067): 0.012*\"pay\" + 0.011*\"money\" + 0.011*\"work\" + 0.010*\"job\" + 0.009*\"year\" + 0.008*\"jobs\" + 0.007*\"people\" + 0.006*\"item\" + 0.005*\"tax\" + 0.005*\"id\"\n",
      "2017-06-05 21:00:22,621 : INFO : topic #10 (0.067): 0.020*\"code\" + 0.012*\"use\" + 0.008*\"language\" + 0.006*\"using\" + 0.005*\"write\" + 0.005*\"systems\" + 0.005*\"python\" + 0.004*\"build\" + 0.004*\"js\" + 0.004*\"rust\"\n",
      "2017-06-05 21:00:22,623 : INFO : topic #3 (0.067): 0.011*\"use\" + 0.008*\"just\" + 0.007*\"user\" + 0.006*\"does\" + 0.006*\"want\" + 0.006*\"phone\" + 0.006*\"desktop\" + 0.005*\"account\" + 0.005*\"information\" + 0.005*\"android\"\n",
      "2017-06-05 21:00:22,625 : INFO : topic #14 (0.067): 0.021*\"data\" + 0.016*\"like\" + 0.009*\"use\" + 0.009*\"type\" + 0.007*\"read\" + 0.007*\"used\" + 0.006*\"case\" + 0.006*\"function\" + 0.005*\"example\" + 0.005*\"looks\"\n",
      "2017-06-05 21:00:22,628 : INFO : topic #8 (0.067): 0.007*\"government\" + 0.007*\"law\" + 0.006*\"state\" + 0.006*\"right\" + 0.006*\"power\" + 0.005*\"people\" + 0.005*\"like\" + 0.005*\"bad\" + 0.005*\"political\" + 0.004*\"states\"\n",
      "2017-06-05 21:00:22,635 : INFO : topic diff=0.048410, rho=0.104257\n",
      "2017-06-05 21:00:22,661 : INFO : PROGRESS: pass 0, at document #186000/246175\n",
      "2017-06-05 21:00:24,180 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:24,435 : INFO : topic #12 (0.067): 0.052*\"com\" + 0.039*\"https\" + 0.035*\"www\" + 0.028*\"http\" + 0.020*\"https www\" + 0.014*\"http www\" + 0.013*\"ago\" + 0.012*\"amazon\" + 0.010*\"years\" + 0.008*\"youtube\"\n",
      "2017-06-05 21:00:24,438 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.009*\"doesn\" + 0.008*\"say\" + 0.007*\"point\" + 0.007*\"just\" + 0.007*\"know\" + 0.006*\"like\" + 0.006*\"mean\"\n",
      "2017-06-05 21:00:24,441 : INFO : topic #8 (0.067): 0.007*\"law\" + 0.007*\"government\" + 0.006*\"state\" + 0.006*\"right\" + 0.005*\"power\" + 0.005*\"people\" + 0.005*\"bad\" + 0.005*\"like\" + 0.005*\"political\" + 0.004*\"states\"\n",
      "2017-06-05 21:00:24,443 : INFO : topic #3 (0.067): 0.011*\"use\" + 0.008*\"just\" + 0.007*\"user\" + 0.006*\"does\" + 0.006*\"desktop\" + 0.006*\"want\" + 0.006*\"phone\" + 0.005*\"account\" + 0.005*\"information\" + 0.005*\"android\"\n",
      "2017-06-05 21:00:24,445 : INFO : topic #9 (0.067): 0.012*\"pay\" + 0.011*\"money\" + 0.011*\"work\" + 0.010*\"job\" + 0.009*\"year\" + 0.008*\"jobs\" + 0.007*\"people\" + 0.006*\"item\" + 0.005*\"id\" + 0.005*\"news\"\n",
      "2017-06-05 21:00:24,452 : INFO : topic diff=0.046490, rho=0.103695\n",
      "2017-06-05 21:00:24,479 : INFO : PROGRESS: pass 0, at document #188000/246175\n",
      "2017-06-05 21:00:25,895 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:26,125 : INFO : topic #12 (0.067): 0.052*\"com\" + 0.040*\"https\" + 0.036*\"www\" + 0.029*\"http\" + 0.020*\"https www\" + 0.014*\"http www\" + 0.013*\"ago\" + 0.011*\"amazon\" + 0.010*\"years\" + 0.008*\"youtube\"\n",
      "2017-06-05 21:00:26,127 : INFO : topic #2 (0.067): 0.021*\"google\" + 0.012*\"open\" + 0.009*\"angular\" + 0.009*\"software\" + 0.008*\"users\" + 0.008*\"don\" + 0.008*\"product\" + 0.008*\"use\" + 0.008*\"source\" + 0.008*\"content\"\n",
      "2017-06-05 21:00:26,129 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.007*\"hn\" + 0.007*\"people\" + 0.006*\"like\" + 0.005*\"ve\" + 0.005*\"laptop\" + 0.005*\"social\" + 0.004*\"didn\" + 0.004*\"just\"\n",
      "2017-06-05 21:00:26,132 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.009*\"doesn\" + 0.008*\"say\" + 0.007*\"point\" + 0.007*\"just\" + 0.007*\"know\" + 0.006*\"like\" + 0.006*\"mean\"\n",
      "2017-06-05 21:00:26,135 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.015*\"org\" + 0.011*\"wikipedia\" + 0.010*\"windows\" + 0.010*\"apps\" + 0.010*\"en\" + 0.010*\"web\" + 0.010*\"app\" + 0.009*\"wiki\" + 0.009*\"github\"\n",
      "2017-06-05 21:00:26,142 : INFO : topic diff=0.045929, rho=0.103142\n",
      "2017-06-05 21:00:26,170 : INFO : PROGRESS: pass 0, at document #190000/246175\n",
      "2017-06-05 21:00:27,714 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:27,948 : INFO : topic #10 (0.067): 0.020*\"code\" + 0.013*\"use\" + 0.008*\"language\" + 0.006*\"using\" + 0.006*\"js\" + 0.005*\"javascript\" + 0.005*\"write\" + 0.004*\"build\" + 0.004*\"systems\" + 0.004*\"python\"\n",
      "2017-06-05 21:00:27,950 : INFO : topic #7 (0.067): 0.007*\"did\" + 0.007*\"women\" + 0.007*\"history\" + 0.006*\"war\" + 0.005*\"chinese\" + 0.004*\"say\" + 0.004*\"china\" + 0.004*\"author\" + 0.004*\"read\" + 0.004*\"wasn\"\n",
      "2017-06-05 21:00:27,953 : INFO : topic #2 (0.067): 0.021*\"google\" + 0.011*\"open\" + 0.009*\"users\" + 0.009*\"don\" + 0.008*\"software\" + 0.008*\"use\" + 0.008*\"angular\" + 0.008*\"source\" + 0.008*\"product\" + 0.008*\"content\"\n",
      "2017-06-05 21:00:27,955 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.007*\"user\" + 0.006*\"does\" + 0.006*\"want\" + 0.006*\"desktop\" + 0.006*\"phone\" + 0.005*\"password\" + 0.005*\"account\" + 0.005*\"information\"\n",
      "2017-06-05 21:00:27,957 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.007*\"hn\" + 0.007*\"people\" + 0.006*\"like\" + 0.005*\"ve\" + 0.005*\"laptop\" + 0.005*\"privacy\" + 0.004*\"social\" + 0.004*\"didn\"\n",
      "2017-06-05 21:00:27,964 : INFO : topic diff=0.043890, rho=0.102598\n",
      "2017-06-05 21:00:27,992 : INFO : PROGRESS: pass 0, at document #192000/246175\n",
      "2017-06-05 21:00:29,521 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:29,748 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.007*\"post\" + 0.007*\"hn\" + 0.007*\"people\" + 0.006*\"like\" + 0.005*\"laptop\" + 0.005*\"ve\" + 0.005*\"battery\" + 0.004*\"privacy\" + 0.004*\"just\"\n",
      "2017-06-05 21:00:29,750 : INFO : topic #2 (0.067): 0.021*\"google\" + 0.011*\"open\" + 0.009*\"users\" + 0.009*\"don\" + 0.008*\"use\" + 0.008*\"software\" + 0.008*\"content\" + 0.008*\"source\" + 0.008*\"product\" + 0.007*\"make\"\n",
      "2017-06-05 21:00:29,753 : INFO : topic #6 (0.067): 0.020*\"don\" + 0.019*\"just\" + 0.017*\"think\" + 0.015*\"like\" + 0.014*\"really\" + 0.013*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.008*\"know\" + 0.008*\"way\"\n",
      "2017-06-05 21:00:29,755 : INFO : topic #0 (0.067): 0.009*\"market\" + 0.008*\"business\" + 0.008*\"uber\" + 0.008*\"companies\" + 0.007*\"cost\" + 0.007*\"company\" + 0.007*\"people\" + 0.006*\"price\" + 0.006*\"tech\" + 0.005*\"car\"\n",
      "2017-06-05 21:00:29,758 : INFO : topic #14 (0.067): 0.020*\"data\" + 0.017*\"like\" + 0.009*\"use\" + 0.008*\"type\" + 0.007*\"used\" + 0.006*\"case\" + 0.006*\"read\" + 0.006*\"function\" + 0.005*\"looks\" + 0.005*\"example\"\n",
      "2017-06-05 21:00:29,766 : INFO : topic diff=0.043726, rho=0.102062\n",
      "2017-06-05 21:00:29,793 : INFO : PROGRESS: pass 0, at document #194000/246175\n",
      "2017-06-05 21:00:31,414 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:31,638 : INFO : topic #7 (0.067): 0.007*\"did\" + 0.007*\"history\" + 0.006*\"women\" + 0.006*\"war\" + 0.005*\"chinese\" + 0.004*\"author\" + 0.004*\"coffee\" + 0.004*\"say\" + 0.004*\"china\" + 0.004*\"reddit\"\n",
      "2017-06-05 21:00:31,640 : INFO : topic #12 (0.067): 0.052*\"com\" + 0.040*\"https\" + 0.034*\"www\" + 0.028*\"http\" + 0.020*\"https www\" + 0.014*\"http www\" + 0.013*\"ago\" + 0.010*\"years\" + 0.009*\"amazon\" + 0.008*\"youtube\"\n",
      "2017-06-05 21:00:31,643 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.007*\"user\" + 0.006*\"does\" + 0.006*\"want\" + 0.006*\"desktop\" + 0.005*\"phone\" + 0.005*\"account\" + 0.005*\"information\" + 0.005*\"access\"\n",
      "2017-06-05 21:00:31,645 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"hn\" + 0.007*\"post\" + 0.007*\"people\" + 0.006*\"like\" + 0.005*\"ve\" + 0.005*\"laptop\" + 0.005*\"social\" + 0.004*\"didn\" + 0.004*\"battery\"\n",
      "2017-06-05 21:00:31,647 : INFO : topic #0 (0.067): 0.009*\"market\" + 0.008*\"business\" + 0.008*\"uber\" + 0.008*\"companies\" + 0.007*\"cost\" + 0.007*\"people\" + 0.007*\"company\" + 0.006*\"price\" + 0.005*\"tech\" + 0.005*\"car\"\n",
      "2017-06-05 21:00:31,654 : INFO : topic diff=0.041966, rho=0.101535\n",
      "2017-06-05 21:00:31,679 : INFO : PROGRESS: pass 0, at document #196000/246175\n",
      "2017-06-05 21:00:33,114 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:33,339 : INFO : topic #6 (0.067): 0.020*\"don\" + 0.019*\"just\" + 0.017*\"think\" + 0.016*\"like\" + 0.014*\"really\" + 0.012*\"people\" + 0.012*\"work\" + 0.009*\"need\" + 0.008*\"know\" + 0.008*\"time\"\n",
      "2017-06-05 21:00:33,341 : INFO : topic #13 (0.067): 0.014*\"problem\" + 0.013*\"time\" + 0.011*\"good\" + 0.009*\"things\" + 0.008*\"just\" + 0.007*\"ve\" + 0.006*\"lot\" + 0.006*\"problems\" + 0.006*\"better\" + 0.006*\"like\"\n",
      "2017-06-05 21:00:33,344 : INFO : topic #10 (0.067): 0.021*\"code\" + 0.012*\"use\" + 0.007*\"language\" + 0.006*\"using\" + 0.006*\"js\" + 0.005*\"javascript\" + 0.005*\"write\" + 0.004*\"build\" + 0.004*\"memory\" + 0.004*\"machine\"\n",
      "2017-06-05 21:00:33,346 : INFO : topic #12 (0.067): 0.051*\"com\" + 0.039*\"https\" + 0.034*\"www\" + 0.028*\"http\" + 0.020*\"https www\" + 0.014*\"http www\" + 0.013*\"ago\" + 0.010*\"years\" + 0.009*\"watch\" + 0.009*\"amazon\"\n",
      "2017-06-05 21:00:33,348 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.007*\"user\" + 0.006*\"does\" + 0.006*\"want\" + 0.006*\"desktop\" + 0.005*\"account\" + 0.005*\"netflix\" + 0.005*\"phone\" + 0.005*\"access\"\n",
      "2017-06-05 21:00:33,355 : INFO : topic diff=0.043676, rho=0.101015\n",
      "2017-06-05 21:00:33,380 : INFO : PROGRESS: pass 0, at document #198000/246175\n",
      "2017-06-05 21:00:34,853 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:35,070 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.013*\"don\" + 0.009*\"doesn\" + 0.009*\"think\" + 0.008*\"say\" + 0.007*\"point\" + 0.007*\"just\" + 0.007*\"know\" + 0.006*\"mean\" + 0.006*\"like\"\n",
      "2017-06-05 21:00:35,072 : INFO : topic #14 (0.067): 0.019*\"data\" + 0.017*\"like\" + 0.009*\"use\" + 0.008*\"type\" + 0.007*\"used\" + 0.007*\"case\" + 0.006*\"read\" + 0.006*\"looks\" + 0.005*\"example\" + 0.005*\"function\"\n",
      "2017-06-05 21:00:35,075 : INFO : topic #13 (0.067): 0.014*\"problem\" + 0.014*\"time\" + 0.011*\"good\" + 0.009*\"things\" + 0.008*\"just\" + 0.007*\"ve\" + 0.006*\"lot\" + 0.006*\"problems\" + 0.006*\"better\" + 0.006*\"like\"\n",
      "2017-06-05 21:00:35,077 : INFO : topic #0 (0.067): 0.010*\"market\" + 0.009*\"business\" + 0.008*\"cost\" + 0.007*\"companies\" + 0.007*\"price\" + 0.007*\"uber\" + 0.007*\"company\" + 0.006*\"people\" + 0.005*\"tech\" + 0.005*\"long\"\n",
      "2017-06-05 21:00:35,079 : INFO : topic #5 (0.067): 0.021*\"https\" + 0.015*\"org\" + 0.012*\"web\" + 0.011*\"windows\" + 0.011*\"browser\" + 0.009*\"wikipedia\" + 0.009*\"en\" + 0.009*\"use\" + 0.009*\"apps\" + 0.008*\"wiki\"\n",
      "2017-06-05 21:00:35,086 : INFO : topic diff=0.042630, rho=0.100504\n",
      "2017-06-05 21:00:38,790 : INFO : -9.510 per-word bound, 729.3 perplexity estimate based on a held-out corpus of 2000 documents with 62771 words\n",
      "2017-06-05 21:00:38,791 : INFO : PROGRESS: pass 0, at document #200000/246175\n",
      "2017-06-05 21:00:40,238 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:40,460 : INFO : topic #2 (0.067): 0.018*\"google\" + 0.013*\"open\" + 0.013*\"content\" + 0.009*\"don\" + 0.009*\"users\" + 0.008*\"use\" + 0.008*\"product\" + 0.008*\"software\" + 0.008*\"drm\" + 0.008*\"free\"\n",
      "2017-06-05 21:00:40,461 : INFO : topic #12 (0.067): 0.052*\"com\" + 0.040*\"https\" + 0.034*\"www\" + 0.029*\"http\" + 0.019*\"https www\" + 0.014*\"http www\" + 0.012*\"ago\" + 0.010*\"years\" + 0.008*\"watch\" + 0.008*\"video\"\n",
      "2017-06-05 21:00:40,464 : INFO : topic #14 (0.067): 0.020*\"data\" + 0.017*\"like\" + 0.009*\"use\" + 0.008*\"type\" + 0.007*\"used\" + 0.007*\"case\" + 0.006*\"read\" + 0.006*\"looks\" + 0.005*\"example\" + 0.005*\"function\"\n",
      "2017-06-05 21:00:40,466 : INFO : topic #11 (0.067): 0.007*\"water\" + 0.006*\"energy\" + 0.005*\"power\" + 0.004*\"drug\" + 0.004*\"city\" + 0.004*\"food\" + 0.004*\"world\" + 0.003*\"large\" + 0.003*\"new\" + 0.003*\"space\"\n",
      "2017-06-05 21:00:40,468 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.007*\"user\" + 0.006*\"want\" + 0.006*\"does\" + 0.006*\"desktop\" + 0.006*\"account\" + 0.005*\"netflix\" + 0.005*\"phone\" + 0.005*\"access\"\n",
      "2017-06-05 21:00:40,475 : INFO : topic diff=0.038491, rho=0.100000\n",
      "2017-06-05 21:00:40,500 : INFO : PROGRESS: pass 0, at document #202000/246175\n",
      "2017-06-05 21:00:41,944 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:42,167 : INFO : topic #11 (0.067): 0.006*\"water\" + 0.006*\"energy\" + 0.005*\"power\" + 0.004*\"drug\" + 0.004*\"city\" + 0.004*\"police\" + 0.004*\"world\" + 0.003*\"food\" + 0.003*\"space\" + 0.003*\"large\"\n",
      "2017-06-05 21:00:42,169 : INFO : topic #5 (0.067): 0.021*\"https\" + 0.015*\"org\" + 0.011*\"web\" + 0.011*\"windows\" + 0.010*\"browser\" + 0.009*\"wikipedia\" + 0.009*\"use\" + 0.009*\"apps\" + 0.009*\"en\" + 0.008*\"wiki\"\n",
      "2017-06-05 21:00:42,171 : INFO : topic #0 (0.067): 0.010*\"market\" + 0.009*\"business\" + 0.008*\"companies\" + 0.008*\"cost\" + 0.007*\"price\" + 0.007*\"company\" + 0.006*\"people\" + 0.006*\"uber\" + 0.005*\"car\" + 0.005*\"amazon\"\n",
      "2017-06-05 21:00:42,173 : INFO : topic #7 (0.067): 0.008*\"coffee\" + 0.007*\"did\" + 0.006*\"history\" + 0.005*\"women\" + 0.005*\"war\" + 0.005*\"patent\" + 0.004*\"author\" + 0.004*\"later\" + 0.004*\"say\" + 0.004*\"wasn\"\n",
      "2017-06-05 21:00:42,175 : INFO : topic #14 (0.067): 0.020*\"data\" + 0.017*\"like\" + 0.009*\"use\" + 0.008*\"type\" + 0.007*\"used\" + 0.007*\"case\" + 0.006*\"read\" + 0.006*\"looks\" + 0.005*\"example\" + 0.005*\"function\"\n",
      "2017-06-05 21:00:42,182 : INFO : topic diff=0.037358, rho=0.099504\n",
      "2017-06-05 21:00:42,207 : INFO : PROGRESS: pass 0, at document #204000/246175\n",
      "2017-06-05 21:00:43,634 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:43,847 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"doesn\" + 0.009*\"think\" + 0.008*\"say\" + 0.007*\"just\" + 0.007*\"point\" + 0.007*\"know\" + 0.006*\"like\" + 0.006*\"mean\"\n",
      "2017-06-05 21:00:43,849 : INFO : topic #5 (0.067): 0.021*\"https\" + 0.015*\"org\" + 0.011*\"web\" + 0.010*\"windows\" + 0.010*\"browser\" + 0.009*\"wikipedia\" + 0.009*\"use\" + 0.009*\"apps\" + 0.009*\"en\" + 0.009*\"github\"\n",
      "2017-06-05 21:00:43,851 : INFO : topic #10 (0.067): 0.021*\"code\" + 0.012*\"use\" + 0.007*\"using\" + 0.007*\"language\" + 0.005*\"js\" + 0.005*\"performance\" + 0.005*\"write\" + 0.005*\"javascript\" + 0.004*\"memory\" + 0.004*\"machine\"\n",
      "2017-06-05 21:00:43,854 : INFO : topic #8 (0.067): 0.008*\"law\" + 0.007*\"government\" + 0.007*\"state\" + 0.006*\"right\" + 0.005*\"power\" + 0.005*\"people\" + 0.005*\"like\" + 0.005*\"bad\" + 0.004*\"political\" + 0.004*\"states\"\n",
      "2017-06-05 21:00:43,856 : INFO : topic #14 (0.067): 0.020*\"data\" + 0.016*\"like\" + 0.009*\"use\" + 0.008*\"type\" + 0.007*\"case\" + 0.007*\"used\" + 0.006*\"read\" + 0.006*\"looks\" + 0.005*\"example\" + 0.005*\"function\"\n",
      "2017-06-05 21:00:43,863 : INFO : topic diff=0.040043, rho=0.099015\n",
      "2017-06-05 21:00:43,887 : INFO : PROGRESS: pass 0, at document #206000/246175\n",
      "2017-06-05 21:00:45,385 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:45,602 : INFO : topic #7 (0.067): 0.008*\"did\" + 0.007*\"coffee\" + 0.005*\"history\" + 0.005*\"women\" + 0.005*\"seller\" + 0.005*\"war\" + 0.004*\"later\" + 0.004*\"patent\" + 0.004*\"author\" + 0.004*\"wasn\"\n",
      "2017-06-05 21:00:45,605 : INFO : topic #13 (0.067): 0.015*\"time\" + 0.015*\"problem\" + 0.011*\"good\" + 0.009*\"things\" + 0.008*\"just\" + 0.008*\"ve\" + 0.007*\"lot\" + 0.006*\"problems\" + 0.006*\"years\" + 0.006*\"better\"\n",
      "2017-06-05 21:00:45,608 : INFO : topic #0 (0.067): 0.010*\"business\" + 0.010*\"market\" + 0.008*\"companies\" + 0.008*\"cost\" + 0.008*\"price\" + 0.008*\"company\" + 0.006*\"people\" + 0.006*\"amazon\" + 0.005*\"uber\" + 0.005*\"buy\"\n",
      "2017-06-05 21:00:45,611 : INFO : topic #2 (0.067): 0.017*\"google\" + 0.012*\"open\" + 0.011*\"content\" + 0.009*\"product\" + 0.009*\"don\" + 0.009*\"free\" + 0.008*\"use\" + 0.008*\"software\" + 0.008*\"users\" + 0.008*\"make\"\n",
      "2017-06-05 21:00:45,613 : INFO : topic #12 (0.067): 0.051*\"com\" + 0.039*\"https\" + 0.033*\"www\" + 0.028*\"http\" + 0.018*\"https www\" + 0.014*\"http www\" + 0.012*\"ago\" + 0.010*\"years\" + 0.009*\"amazon\" + 0.009*\"apple\"\n",
      "2017-06-05 21:00:45,620 : INFO : topic diff=0.037874, rho=0.098533\n",
      "2017-06-05 21:00:45,645 : INFO : PROGRESS: pass 0, at document #208000/246175\n",
      "2017-06-05 21:00:47,043 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:47,251 : INFO : topic #12 (0.067): 0.051*\"com\" + 0.039*\"https\" + 0.033*\"www\" + 0.028*\"http\" + 0.019*\"https www\" + 0.013*\"http www\" + 0.012*\"ago\" + 0.010*\"ipad\" + 0.010*\"years\" + 0.010*\"apple\"\n",
      "2017-06-05 21:00:47,254 : INFO : topic #13 (0.067): 0.014*\"time\" + 0.014*\"problem\" + 0.011*\"good\" + 0.009*\"things\" + 0.008*\"ve\" + 0.008*\"just\" + 0.007*\"lot\" + 0.006*\"problems\" + 0.006*\"years\" + 0.006*\"better\"\n",
      "2017-06-05 21:00:47,256 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.007*\"post\" + 0.007*\"pro\" + 0.007*\"hn\" + 0.006*\"like\" + 0.006*\"laptop\" + 0.006*\"people\" + 0.005*\"ve\" + 0.005*\"battery\" + 0.004*\"social\"\n",
      "2017-06-05 21:00:47,258 : INFO : topic #9 (0.067): 0.013*\"work\" + 0.013*\"pay\" + 0.012*\"money\" + 0.009*\"job\" + 0.008*\"year\" + 0.008*\"jobs\" + 0.008*\"people\" + 0.007*\"tax\" + 0.006*\"company\" + 0.006*\"item\"\n",
      "2017-06-05 21:00:47,260 : INFO : topic #0 (0.067): 0.010*\"business\" + 0.010*\"market\" + 0.008*\"companies\" + 0.008*\"company\" + 0.008*\"price\" + 0.008*\"cost\" + 0.006*\"people\" + 0.006*\"amazon\" + 0.005*\"uber\" + 0.005*\"car\"\n",
      "2017-06-05 21:00:47,267 : INFO : topic diff=0.037417, rho=0.098058\n",
      "2017-06-05 21:00:47,294 : INFO : PROGRESS: pass 0, at document #210000/246175\n",
      "2017-06-05 21:00:48,689 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:48,896 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.007*\"password\" + 0.007*\"user\" + 0.006*\"phone\" + 0.006*\"does\" + 0.006*\"want\" + 0.006*\"desktop\" + 0.006*\"android\" + 0.005*\"account\"\n",
      "2017-06-05 21:00:48,898 : INFO : topic #10 (0.067): 0.021*\"code\" + 0.012*\"use\" + 0.007*\"language\" + 0.007*\"using\" + 0.005*\"write\" + 0.005*\"machine\" + 0.004*\"js\" + 0.004*\"memory\" + 0.004*\"time\" + 0.004*\"systems\"\n",
      "2017-06-05 21:00:48,900 : INFO : topic #4 (0.067): 0.010*\"life\" + 0.008*\"post\" + 0.007*\"hn\" + 0.007*\"laptop\" + 0.006*\"like\" + 0.006*\"pro\" + 0.006*\"people\" + 0.005*\"ve\" + 0.005*\"battery\" + 0.004*\"title\"\n",
      "2017-06-05 21:00:48,902 : INFO : topic #13 (0.067): 0.014*\"time\" + 0.014*\"problem\" + 0.011*\"good\" + 0.009*\"things\" + 0.008*\"ve\" + 0.007*\"just\" + 0.007*\"lot\" + 0.006*\"problems\" + 0.006*\"years\" + 0.006*\"better\"\n",
      "2017-06-05 21:00:48,904 : INFO : topic #9 (0.067): 0.013*\"work\" + 0.013*\"pay\" + 0.012*\"money\" + 0.009*\"job\" + 0.008*\"year\" + 0.008*\"tax\" + 0.008*\"people\" + 0.007*\"jobs\" + 0.006*\"company\" + 0.006*\"hours\"\n",
      "2017-06-05 21:00:48,911 : INFO : topic diff=0.040748, rho=0.097590\n",
      "2017-06-05 21:00:48,937 : INFO : PROGRESS: pass 0, at document #212000/246175\n",
      "2017-06-05 21:00:50,369 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:50,574 : INFO : topic #6 (0.067): 0.020*\"don\" + 0.018*\"just\" + 0.017*\"think\" + 0.016*\"like\" + 0.014*\"really\" + 0.012*\"people\" + 0.012*\"work\" + 0.009*\"need\" + 0.009*\"way\" + 0.009*\"know\"\n",
      "2017-06-05 21:00:50,576 : INFO : topic #7 (0.067): 0.009*\"art\" + 0.008*\"did\" + 0.006*\"history\" + 0.005*\"coffee\" + 0.005*\"war\" + 0.005*\"author\" + 0.004*\"wasn\" + 0.004*\"patent\" + 0.004*\"later\" + 0.004*\"women\"\n",
      "2017-06-05 21:00:50,579 : INFO : topic #8 (0.067): 0.008*\"law\" + 0.008*\"government\" + 0.007*\"state\" + 0.006*\"right\" + 0.005*\"people\" + 0.005*\"like\" + 0.005*\"power\" + 0.005*\"states\" + 0.005*\"bad\" + 0.005*\"legal\"\n",
      "2017-06-05 21:00:50,581 : INFO : topic #10 (0.067): 0.020*\"code\" + 0.012*\"use\" + 0.007*\"language\" + 0.007*\"using\" + 0.006*\"write\" + 0.005*\"machine\" + 0.004*\"memory\" + 0.004*\"js\" + 0.004*\"time\" + 0.004*\"run\"\n",
      "2017-06-05 21:00:50,583 : INFO : topic #1 (0.067): 0.024*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.009*\"doesn\" + 0.008*\"say\" + 0.007*\"know\" + 0.007*\"point\" + 0.007*\"just\" + 0.006*\"mean\" + 0.006*\"like\"\n",
      "2017-06-05 21:00:50,589 : INFO : topic diff=0.043364, rho=0.097129\n",
      "2017-06-05 21:00:50,614 : INFO : PROGRESS: pass 0, at document #214000/246175\n",
      "2017-06-05 21:00:52,137 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:52,355 : INFO : topic #9 (0.067): 0.013*\"money\" + 0.013*\"pay\" + 0.012*\"work\" + 0.008*\"job\" + 0.008*\"people\" + 0.008*\"year\" + 0.008*\"tax\" + 0.008*\"insurance\" + 0.007*\"jobs\" + 0.006*\"uk\"\n",
      "2017-06-05 21:00:52,358 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.006*\"phone\" + 0.006*\"user\" + 0.006*\"does\" + 0.006*\"password\" + 0.006*\"want\" + 0.005*\"desktop\" + 0.005*\"information\" + 0.005*\"android\"\n",
      "2017-06-05 21:00:52,360 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.007*\"post\" + 0.006*\"like\" + 0.006*\"hn\" + 0.006*\"people\" + 0.006*\"pro\" + 0.006*\"laptop\" + 0.005*\"ve\" + 0.005*\"battery\" + 0.004*\"didn\"\n",
      "2017-06-05 21:00:52,362 : INFO : topic #5 (0.067): 0.022*\"https\" + 0.015*\"org\" + 0.010*\"github\" + 0.010*\"wikipedia\" + 0.010*\"en\" + 0.010*\"windows\" + 0.009*\"wiki\" + 0.009*\"web\" + 0.009*\"use\" + 0.008*\"org wiki\"\n",
      "2017-06-05 21:00:52,364 : INFO : topic #7 (0.067): 0.008*\"art\" + 0.008*\"did\" + 0.005*\"history\" + 0.005*\"patent\" + 0.005*\"war\" + 0.005*\"coffee\" + 0.004*\"author\" + 0.004*\"patents\" + 0.004*\"wasn\" + 0.004*\"years\"\n",
      "2017-06-05 21:00:52,371 : INFO : topic diff=0.044127, rho=0.096674\n",
      "2017-06-05 21:00:52,399 : INFO : PROGRESS: pass 0, at document #216000/246175\n",
      "2017-06-05 21:00:53,944 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:54,187 : INFO : topic #8 (0.067): 0.009*\"law\" + 0.007*\"government\" + 0.006*\"state\" + 0.006*\"right\" + 0.005*\"people\" + 0.005*\"power\" + 0.005*\"states\" + 0.005*\"bad\" + 0.005*\"like\" + 0.005*\"political\"\n",
      "2017-06-05 21:00:54,189 : INFO : topic #6 (0.067): 0.020*\"don\" + 0.018*\"just\" + 0.018*\"think\" + 0.016*\"like\" + 0.014*\"really\" + 0.012*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.009*\"way\" + 0.009*\"know\"\n",
      "2017-06-05 21:00:54,192 : INFO : topic #7 (0.067): 0.008*\"art\" + 0.008*\"did\" + 0.008*\"patent\" + 0.007*\"patents\" + 0.005*\"history\" + 0.005*\"author\" + 0.005*\"war\" + 0.004*\"years\" + 0.004*\"coffee\" + 0.004*\"wasn\"\n",
      "2017-06-05 21:00:54,194 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.006*\"hn\" + 0.006*\"like\" + 0.006*\"people\" + 0.006*\"pro\" + 0.005*\"laptop\" + 0.005*\"ve\" + 0.005*\"appliances\" + 0.005*\"battery\"\n",
      "2017-06-05 21:00:54,197 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.006*\"does\" + 0.006*\"user\" + 0.006*\"phone\" + 0.006*\"want\" + 0.005*\"password\" + 0.005*\"information\" + 0.005*\"desktop\" + 0.005*\"account\"\n",
      "2017-06-05 21:00:54,206 : INFO : topic diff=0.040223, rho=0.096225\n",
      "2017-06-05 21:00:54,234 : INFO : PROGRESS: pass 0, at document #218000/246175\n",
      "2017-06-05 21:00:55,812 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:00:56,035 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.007*\"hn\" + 0.006*\"like\" + 0.006*\"people\" + 0.005*\"pro\" + 0.005*\"ve\" + 0.005*\"laptop\" + 0.005*\"battery\" + 0.004*\"title\"\n",
      "2017-06-05 21:00:56,037 : INFO : topic #7 (0.067): 0.008*\"did\" + 0.008*\"art\" + 0.007*\"patent\" + 0.006*\"patents\" + 0.005*\"history\" + 0.005*\"war\" + 0.005*\"author\" + 0.004*\"years\" + 0.004*\"wasn\" + 0.004*\"later\"\n",
      "2017-06-05 21:00:56,039 : INFO : topic #13 (0.067): 0.014*\"time\" + 0.014*\"problem\" + 0.011*\"good\" + 0.009*\"things\" + 0.008*\"ai\" + 0.008*\"ve\" + 0.007*\"just\" + 0.007*\"years\" + 0.007*\"lot\" + 0.007*\"problems\"\n",
      "2017-06-05 21:00:56,042 : INFO : topic #14 (0.067): 0.023*\"data\" + 0.016*\"like\" + 0.009*\"use\" + 0.008*\"type\" + 0.007*\"case\" + 0.007*\"used\" + 0.006*\"looks\" + 0.006*\"read\" + 0.005*\"function\" + 0.005*\"example\"\n",
      "2017-06-05 21:00:56,044 : INFO : topic #11 (0.067): 0.007*\"energy\" + 0.006*\"water\" + 0.005*\"power\" + 0.004*\"new\" + 0.004*\"world\" + 0.004*\"food\" + 0.003*\"air\" + 0.003*\"drug\" + 0.003*\"police\" + 0.003*\"city\"\n",
      "2017-06-05 21:00:56,052 : INFO : topic diff=0.039108, rho=0.095783\n",
      "2017-06-05 21:01:00,479 : INFO : -9.450 per-word bound, 699.6 perplexity estimate based on a held-out corpus of 2000 documents with 78734 words\n",
      "2017-06-05 21:01:00,480 : INFO : PROGRESS: pass 0, at document #220000/246175\n",
      "2017-06-05 21:01:01,996 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:02,204 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.007*\"hn\" + 0.006*\"like\" + 0.006*\"people\" + 0.005*\"ve\" + 0.005*\"pro\" + 0.005*\"laptop\" + 0.004*\"social\" + 0.004*\"battery\"\n",
      "2017-06-05 21:01:02,206 : INFO : topic #14 (0.067): 0.022*\"data\" + 0.016*\"like\" + 0.009*\"use\" + 0.008*\"type\" + 0.007*\"used\" + 0.007*\"case\" + 0.006*\"looks\" + 0.006*\"read\" + 0.005*\"function\" + 0.005*\"example\"\n",
      "2017-06-05 21:01:02,208 : INFO : topic #0 (0.067): 0.009*\"business\" + 0.009*\"market\" + 0.008*\"companies\" + 0.008*\"cost\" + 0.008*\"uber\" + 0.007*\"company\" + 0.007*\"price\" + 0.006*\"people\" + 0.005*\"buy\" + 0.005*\"car\"\n",
      "2017-06-05 21:01:02,210 : INFO : topic #13 (0.067): 0.014*\"time\" + 0.014*\"problem\" + 0.011*\"good\" + 0.009*\"things\" + 0.008*\"ve\" + 0.008*\"ai\" + 0.007*\"just\" + 0.007*\"problems\" + 0.007*\"years\" + 0.007*\"lot\"\n",
      "2017-06-05 21:01:02,212 : INFO : topic #9 (0.067): 0.013*\"money\" + 0.012*\"pay\" + 0.011*\"work\" + 0.009*\"people\" + 0.008*\"job\" + 0.008*\"year\" + 0.008*\"jobs\" + 0.007*\"insurance\" + 0.007*\"tax\" + 0.005*\"country\"\n",
      "2017-06-05 21:01:02,218 : INFO : topic diff=0.039229, rho=0.095346\n",
      "2017-06-05 21:01:02,243 : INFO : PROGRESS: pass 0, at document #222000/246175\n",
      "2017-06-05 21:01:03,703 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:03,901 : INFO : topic #11 (0.067): 0.007*\"energy\" + 0.006*\"water\" + 0.005*\"power\" + 0.004*\"drug\" + 0.004*\"world\" + 0.004*\"new\" + 0.004*\"police\" + 0.004*\"city\" + 0.003*\"food\" + 0.003*\"air\"\n",
      "2017-06-05 21:01:03,903 : INFO : topic #5 (0.067): 0.022*\"https\" + 0.015*\"org\" + 0.015*\"windows\" + 0.012*\"wikipedia\" + 0.011*\"github\" + 0.010*\"en\" + 0.010*\"wiki\" + 0.009*\"use\" + 0.009*\"linux\" + 0.009*\"web\"\n",
      "2017-06-05 21:01:03,905 : INFO : topic #12 (0.067): 0.053*\"com\" + 0.039*\"https\" + 0.036*\"www\" + 0.028*\"http\" + 0.019*\"https www\" + 0.014*\"http www\" + 0.013*\"ago\" + 0.012*\"years\" + 0.009*\"apple\" + 0.008*\"news\"\n",
      "2017-06-05 21:01:03,907 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.006*\"does\" + 0.006*\"want\" + 0.006*\"user\" + 0.006*\"desktop\" + 0.005*\"phone\" + 0.005*\"information\" + 0.005*\"using\" + 0.005*\"access\"\n",
      "2017-06-05 21:01:03,909 : INFO : topic #14 (0.067): 0.022*\"data\" + 0.016*\"like\" + 0.009*\"use\" + 0.008*\"type\" + 0.007*\"used\" + 0.007*\"case\" + 0.006*\"looks\" + 0.006*\"read\" + 0.005*\"example\" + 0.005*\"function\"\n",
      "2017-06-05 21:01:03,917 : INFO : topic diff=0.035958, rho=0.094916\n",
      "2017-06-05 21:01:03,942 : INFO : PROGRESS: pass 0, at document #224000/246175\n",
      "2017-06-05 21:01:05,372 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:05,568 : INFO : topic #5 (0.067): 0.022*\"https\" + 0.015*\"org\" + 0.014*\"windows\" + 0.011*\"wikipedia\" + 0.011*\"github\" + 0.010*\"en\" + 0.009*\"use\" + 0.009*\"wiki\" + 0.009*\"linux\" + 0.009*\"web\"\n",
      "2017-06-05 21:01:05,571 : INFO : topic #13 (0.067): 0.015*\"time\" + 0.014*\"problem\" + 0.011*\"good\" + 0.009*\"things\" + 0.008*\"ve\" + 0.007*\"ai\" + 0.007*\"just\" + 0.007*\"problems\" + 0.007*\"lot\" + 0.007*\"years\"\n",
      "2017-06-05 21:01:05,572 : INFO : topic #14 (0.067): 0.021*\"data\" + 0.016*\"like\" + 0.009*\"use\" + 0.008*\"type\" + 0.007*\"used\" + 0.007*\"case\" + 0.006*\"looks\" + 0.006*\"read\" + 0.005*\"example\" + 0.005*\"function\"\n",
      "2017-06-05 21:01:05,574 : INFO : topic #12 (0.067): 0.054*\"com\" + 0.040*\"https\" + 0.037*\"www\" + 0.029*\"http\" + 0.019*\"https www\" + 0.015*\"http www\" + 0.013*\"ago\" + 0.011*\"years\" + 0.009*\"apple\" + 0.008*\"news\"\n",
      "2017-06-05 21:01:05,576 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"hn\" + 0.008*\"post\" + 0.006*\"like\" + 0.006*\"people\" + 0.006*\"pro\" + 0.005*\"ve\" + 0.005*\"laptop\" + 0.004*\"social\" + 0.004*\"title\"\n",
      "2017-06-05 21:01:05,582 : INFO : topic diff=0.033820, rho=0.094491\n",
      "2017-06-05 21:01:05,606 : INFO : PROGRESS: pass 0, at document #226000/246175\n",
      "2017-06-05 21:01:07,023 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:07,221 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.008*\"hn\" + 0.006*\"like\" + 0.006*\"people\" + 0.005*\"pro\" + 0.005*\"ve\" + 0.005*\"social\" + 0.004*\"title\" + 0.004*\"didn\"\n",
      "2017-06-05 21:01:07,223 : INFO : topic #11 (0.067): 0.008*\"water\" + 0.006*\"energy\" + 0.005*\"city\" + 0.004*\"food\" + 0.004*\"power\" + 0.004*\"world\" + 0.004*\"new\" + 0.003*\"drug\" + 0.003*\"space\" + 0.003*\"police\"\n",
      "2017-06-05 21:01:07,225 : INFO : topic #7 (0.067): 0.008*\"did\" + 0.007*\"art\" + 0.005*\"history\" + 0.005*\"war\" + 0.005*\"japanese\" + 0.005*\"patent\" + 0.004*\"author\" + 0.004*\"patents\" + 0.004*\"later\" + 0.004*\"wasn\"\n",
      "2017-06-05 21:01:07,227 : INFO : topic #12 (0.067): 0.054*\"com\" + 0.040*\"https\" + 0.037*\"www\" + 0.029*\"http\" + 0.019*\"https www\" + 0.015*\"http www\" + 0.013*\"ago\" + 0.011*\"years\" + 0.008*\"apple\" + 0.008*\"news\"\n",
      "2017-06-05 21:01:07,229 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.006*\"does\" + 0.006*\"user\" + 0.006*\"want\" + 0.006*\"desktop\" + 0.005*\"bike\" + 0.005*\"information\" + 0.005*\"phone\" + 0.005*\"access\"\n",
      "2017-06-05 21:01:07,236 : INFO : topic diff=0.032003, rho=0.094072\n",
      "2017-06-05 21:01:07,258 : INFO : PROGRESS: pass 0, at document #228000/246175\n",
      "2017-06-05 21:01:08,723 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:08,916 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.007*\"hn\" + 0.006*\"like\" + 0.006*\"people\" + 0.005*\"ve\" + 0.005*\"pro\" + 0.005*\"social\" + 0.004*\"title\" + 0.004*\"didn\"\n",
      "2017-06-05 21:01:08,918 : INFO : topic #14 (0.067): 0.023*\"data\" + 0.016*\"like\" + 0.009*\"use\" + 0.007*\"type\" + 0.007*\"used\" + 0.007*\"case\" + 0.006*\"looks\" + 0.006*\"read\" + 0.005*\"example\" + 0.005*\"function\"\n",
      "2017-06-05 21:01:08,920 : INFO : topic #12 (0.067): 0.054*\"com\" + 0.040*\"https\" + 0.037*\"www\" + 0.029*\"http\" + 0.019*\"https www\" + 0.015*\"http www\" + 0.013*\"ago\" + 0.011*\"years\" + 0.008*\"apple\" + 0.008*\"news\"\n",
      "2017-06-05 21:01:08,922 : INFO : topic #0 (0.067): 0.009*\"business\" + 0.009*\"market\" + 0.008*\"cost\" + 0.008*\"companies\" + 0.007*\"company\" + 0.007*\"price\" + 0.006*\"people\" + 0.006*\"uber\" + 0.005*\"long\" + 0.005*\"buy\"\n",
      "2017-06-05 21:01:08,925 : INFO : topic #10 (0.067): 0.018*\"code\" + 0.013*\"use\" + 0.007*\"language\" + 0.007*\"using\" + 0.005*\"machine\" + 0.005*\"memory\" + 0.005*\"write\" + 0.004*\"systems\" + 0.004*\"build\" + 0.004*\"used\"\n",
      "2017-06-05 21:01:08,932 : INFO : topic diff=0.038464, rho=0.093659\n",
      "2017-06-05 21:01:08,956 : INFO : PROGRESS: pass 0, at document #230000/246175\n",
      "2017-06-05 21:01:10,424 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:10,616 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.007*\"hn\" + 0.006*\"like\" + 0.006*\"people\" + 0.005*\"ve\" + 0.005*\"social\" + 0.005*\"pro\" + 0.004*\"title\" + 0.004*\"didn\"\n",
      "2017-06-05 21:01:10,618 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"doesn\" + 0.009*\"think\" + 0.008*\"say\" + 0.007*\"point\" + 0.007*\"just\" + 0.007*\"does\" + 0.007*\"mean\" + 0.007*\"know\"\n",
      "2017-06-05 21:01:10,621 : INFO : topic #11 (0.067): 0.008*\"water\" + 0.006*\"energy\" + 0.005*\"city\" + 0.004*\"food\" + 0.004*\"power\" + 0.004*\"new\" + 0.004*\"world\" + 0.003*\"air\" + 0.003*\"drug\" + 0.003*\"space\"\n",
      "2017-06-05 21:01:10,623 : INFO : topic #9 (0.067): 0.013*\"money\" + 0.012*\"pay\" + 0.011*\"work\" + 0.009*\"people\" + 0.008*\"job\" + 0.008*\"jobs\" + 0.007*\"year\" + 0.007*\"tax\" + 0.005*\"item\" + 0.005*\"living\"\n",
      "2017-06-05 21:01:10,625 : INFO : topic #0 (0.067): 0.009*\"business\" + 0.009*\"market\" + 0.008*\"cost\" + 0.008*\"companies\" + 0.007*\"company\" + 0.007*\"price\" + 0.006*\"people\" + 0.005*\"uber\" + 0.005*\"long\" + 0.005*\"buy\"\n",
      "2017-06-05 21:01:10,631 : INFO : topic diff=0.030996, rho=0.093250\n",
      "2017-06-05 21:01:10,653 : INFO : PROGRESS: pass 0, at document #232000/246175\n",
      "2017-06-05 21:01:12,100 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:12,293 : INFO : topic #9 (0.067): 0.013*\"money\" + 0.012*\"pay\" + 0.011*\"work\" + 0.009*\"people\" + 0.009*\"job\" + 0.009*\"jobs\" + 0.007*\"year\" + 0.006*\"tax\" + 0.005*\"item\" + 0.005*\"living\"\n",
      "2017-06-05 21:01:12,295 : INFO : topic #7 (0.067): 0.008*\"did\" + 0.007*\"war\" + 0.007*\"patent\" + 0.006*\"art\" + 0.006*\"history\" + 0.005*\"patents\" + 0.004*\"japanese\" + 0.004*\"author\" + 0.004*\"women\" + 0.004*\"later\"\n",
      "2017-06-05 21:01:12,297 : INFO : topic #13 (0.067): 0.015*\"time\" + 0.013*\"problem\" + 0.011*\"good\" + 0.011*\"game\" + 0.008*\"things\" + 0.008*\"ve\" + 0.008*\"games\" + 0.007*\"just\" + 0.007*\"lot\" + 0.007*\"ai\"\n",
      "2017-06-05 21:01:12,299 : INFO : topic #12 (0.067): 0.052*\"com\" + 0.040*\"https\" + 0.036*\"www\" + 0.028*\"http\" + 0.019*\"https www\" + 0.015*\"http www\" + 0.013*\"ago\" + 0.011*\"years\" + 0.009*\"video\" + 0.009*\"watch\"\n",
      "2017-06-05 21:01:12,301 : INFO : topic #2 (0.067): 0.017*\"google\" + 0.011*\"open\" + 0.011*\"free\" + 0.009*\"don\" + 0.009*\"software\" + 0.009*\"product\" + 0.008*\"users\" + 0.008*\"content\" + 0.008*\"source\" + 0.008*\"use\"\n",
      "2017-06-05 21:01:12,307 : INFO : topic diff=0.030006, rho=0.092848\n",
      "2017-06-05 21:01:12,330 : INFO : PROGRESS: pass 0, at document #234000/246175\n",
      "2017-06-05 21:01:13,831 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:14,022 : INFO : topic #11 (0.067): 0.007*\"water\" + 0.006*\"energy\" + 0.005*\"power\" + 0.005*\"air\" + 0.004*\"city\" + 0.004*\"food\" + 0.004*\"police\" + 0.003*\"world\" + 0.003*\"new\" + 0.003*\"drug\"\n",
      "2017-06-05 21:01:14,024 : INFO : topic #6 (0.067): 0.020*\"don\" + 0.018*\"just\" + 0.017*\"think\" + 0.017*\"like\" + 0.014*\"really\" + 0.012*\"people\" + 0.011*\"work\" + 0.009*\"need\" + 0.009*\"way\" + 0.008*\"time\"\n",
      "2017-06-05 21:01:14,025 : INFO : topic #9 (0.067): 0.013*\"money\" + 0.012*\"pay\" + 0.011*\"work\" + 0.009*\"job\" + 0.009*\"people\" + 0.008*\"jobs\" + 0.007*\"year\" + 0.006*\"tax\" + 0.005*\"item\" + 0.005*\"income\"\n",
      "2017-06-05 21:01:14,028 : INFO : topic #2 (0.067): 0.019*\"google\" + 0.012*\"open\" + 0.010*\"free\" + 0.009*\"software\" + 0.009*\"don\" + 0.009*\"source\" + 0.009*\"product\" + 0.008*\"users\" + 0.008*\"content\" + 0.008*\"use\"\n",
      "2017-06-05 21:01:14,030 : INFO : topic #7 (0.067): 0.008*\"did\" + 0.007*\"war\" + 0.006*\"patent\" + 0.006*\"history\" + 0.006*\"art\" + 0.005*\"patents\" + 0.004*\"author\" + 0.004*\"japanese\" + 0.004*\"later\" + 0.004*\"wasn\"\n",
      "2017-06-05 21:01:14,036 : INFO : topic diff=0.035638, rho=0.092450\n",
      "2017-06-05 21:01:14,060 : INFO : PROGRESS: pass 0, at document #236000/246175\n",
      "2017-06-05 21:01:15,521 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:15,716 : INFO : topic #7 (0.067): 0.008*\"did\" + 0.007*\"war\" + 0.007*\"history\" + 0.006*\"patent\" + 0.005*\"art\" + 0.004*\"patents\" + 0.004*\"later\" + 0.004*\"author\" + 0.004*\"japanese\" + 0.004*\"wasn\"\n",
      "2017-06-05 21:01:15,718 : INFO : topic #12 (0.067): 0.051*\"com\" + 0.039*\"https\" + 0.035*\"www\" + 0.028*\"http\" + 0.019*\"https www\" + 0.015*\"http www\" + 0.012*\"ago\" + 0.011*\"years\" + 0.009*\"video\" + 0.008*\"news\"\n",
      "2017-06-05 21:01:15,720 : INFO : topic #3 (0.067): 0.013*\"use\" + 0.008*\"just\" + 0.007*\"information\" + 0.006*\"user\" + 0.006*\"want\" + 0.006*\"does\" + 0.006*\"access\" + 0.005*\"phone\" + 0.004*\"using\" + 0.004*\"desktop\"\n",
      "2017-06-05 21:01:15,722 : INFO : topic #14 (0.067): 0.025*\"data\" + 0.016*\"like\" + 0.010*\"use\" + 0.007*\"type\" + 0.007*\"used\" + 0.006*\"case\" + 0.006*\"read\" + 0.006*\"looks\" + 0.005*\"example\" + 0.005*\"function\"\n",
      "2017-06-05 21:01:15,724 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.011*\"don\" + 0.009*\"doesn\" + 0.009*\"think\" + 0.008*\"say\" + 0.007*\"point\" + 0.007*\"just\" + 0.007*\"know\" + 0.007*\"does\" + 0.006*\"mean\"\n",
      "2017-06-05 21:01:15,731 : INFO : topic diff=0.031186, rho=0.092057\n",
      "2017-06-05 21:01:15,754 : INFO : PROGRESS: pass 0, at document #238000/246175\n",
      "2017-06-05 21:01:17,178 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:17,376 : INFO : topic #7 (0.067): 0.008*\"did\" + 0.007*\"war\" + 0.007*\"history\" + 0.006*\"patent\" + 0.005*\"art\" + 0.004*\"japanese\" + 0.004*\"author\" + 0.004*\"later\" + 0.004*\"patents\" + 0.004*\"wasn\"\n",
      "2017-06-05 21:01:17,378 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.016*\"org\" + 0.013*\"windows\" + 0.012*\"wikipedia\" + 0.012*\"github\" + 0.010*\"en\" + 0.010*\"wiki\" + 0.009*\"use\" + 0.009*\"web\" + 0.009*\"org wiki\"\n",
      "2017-06-05 21:01:17,381 : INFO : topic #10 (0.067): 0.019*\"code\" + 0.013*\"use\" + 0.007*\"using\" + 0.007*\"language\" + 0.005*\"write\" + 0.005*\"machine\" + 0.005*\"memory\" + 0.004*\"systems\" + 0.004*\"build\" + 0.004*\"time\"\n",
      "2017-06-05 21:01:17,383 : INFO : topic #11 (0.067): 0.006*\"water\" + 0.006*\"energy\" + 0.005*\"police\" + 0.005*\"power\" + 0.004*\"air\" + 0.004*\"city\" + 0.004*\"food\" + 0.004*\"world\" + 0.003*\"space\" + 0.003*\"new\"\n",
      "2017-06-05 21:01:17,385 : INFO : topic #9 (0.067): 0.014*\"money\" + 0.013*\"pay\" + 0.011*\"work\" + 0.009*\"job\" + 0.009*\"people\" + 0.008*\"jobs\" + 0.007*\"year\" + 0.007*\"tax\" + 0.005*\"income\" + 0.005*\"living\"\n",
      "2017-06-05 21:01:17,392 : INFO : topic diff=0.030586, rho=0.091670\n",
      "2017-06-05 21:01:21,595 : INFO : -9.430 per-word bound, 689.7 perplexity estimate based on a held-out corpus of 2000 documents with 69877 words\n",
      "2017-06-05 21:01:21,596 : INFO : PROGRESS: pass 0, at document #240000/246175\n",
      "2017-06-05 21:01:23,055 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:23,243 : INFO : topic #13 (0.067): 0.015*\"time\" + 0.014*\"problem\" + 0.011*\"good\" + 0.009*\"ve\" + 0.009*\"game\" + 0.009*\"things\" + 0.007*\"just\" + 0.007*\"problems\" + 0.007*\"lot\" + 0.006*\"better\"\n",
      "2017-06-05 21:01:23,246 : INFO : topic #10 (0.067): 0.019*\"code\" + 0.014*\"use\" + 0.007*\"language\" + 0.007*\"using\" + 0.005*\"write\" + 0.005*\"memory\" + 0.004*\"machine\" + 0.004*\"systems\" + 0.004*\"java\" + 0.004*\"run\"\n",
      "2017-06-05 21:01:23,248 : INFO : topic #2 (0.067): 0.020*\"google\" + 0.011*\"open\" + 0.010*\"free\" + 0.009*\"don\" + 0.009*\"content\" + 0.009*\"source\" + 0.008*\"software\" + 0.008*\"make\" + 0.008*\"product\" + 0.008*\"use\"\n",
      "2017-06-05 21:01:23,251 : INFO : topic #11 (0.067): 0.006*\"water\" + 0.006*\"energy\" + 0.005*\"police\" + 0.005*\"power\" + 0.004*\"air\" + 0.004*\"food\" + 0.004*\"city\" + 0.004*\"world\" + 0.003*\"space\" + 0.003*\"new\"\n",
      "2017-06-05 21:01:23,253 : INFO : topic #12 (0.067): 0.052*\"com\" + 0.039*\"https\" + 0.035*\"www\" + 0.029*\"http\" + 0.018*\"https www\" + 0.015*\"http www\" + 0.013*\"ago\" + 0.011*\"years\" + 0.009*\"video\" + 0.008*\"watch\"\n",
      "2017-06-05 21:01:23,260 : INFO : topic diff=0.032567, rho=0.091287\n",
      "2017-06-05 21:01:23,282 : INFO : PROGRESS: pass 0, at document #242000/246175\n",
      "2017-06-05 21:01:24,757 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:24,945 : INFO : topic #8 (0.067): 0.009*\"law\" + 0.008*\"government\" + 0.006*\"right\" + 0.006*\"state\" + 0.006*\"people\" + 0.005*\"trump\" + 0.005*\"bad\" + 0.005*\"power\" + 0.005*\"like\" + 0.004*\"states\"\n",
      "2017-06-05 21:01:24,947 : INFO : topic #0 (0.067): 0.009*\"uber\" + 0.008*\"market\" + 0.008*\"cost\" + 0.008*\"business\" + 0.008*\"company\" + 0.007*\"companies\" + 0.007*\"price\" + 0.007*\"car\" + 0.006*\"people\" + 0.005*\"tip\"\n",
      "2017-06-05 21:01:24,949 : INFO : topic #2 (0.067): 0.020*\"google\" + 0.011*\"open\" + 0.010*\"free\" + 0.009*\"content\" + 0.009*\"don\" + 0.008*\"source\" + 0.008*\"product\" + 0.008*\"software\" + 0.008*\"users\" + 0.008*\"make\"\n",
      "2017-06-05 21:01:24,951 : INFO : topic #1 (0.067): 0.026*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.009*\"doesn\" + 0.008*\"say\" + 0.007*\"point\" + 0.007*\"just\" + 0.007*\"know\" + 0.007*\"mean\" + 0.006*\"does\"\n",
      "2017-06-05 21:01:24,953 : INFO : topic #6 (0.067): 0.020*\"don\" + 0.018*\"just\" + 0.018*\"like\" + 0.017*\"think\" + 0.015*\"really\" + 0.012*\"people\" + 0.010*\"work\" + 0.009*\"need\" + 0.009*\"way\" + 0.008*\"know\"\n",
      "2017-06-05 21:01:24,959 : INFO : topic diff=0.033995, rho=0.090909\n",
      "2017-06-05 21:01:24,982 : INFO : PROGRESS: pass 0, at document #244000/246175\n",
      "2017-06-05 21:01:26,440 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:26,644 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.007*\"information\" + 0.006*\"user\" + 0.006*\"want\" + 0.006*\"does\" + 0.005*\"access\" + 0.005*\"phone\" + 0.005*\"using\" + 0.005*\"service\"\n",
      "2017-06-05 21:01:26,646 : INFO : topic #9 (0.067): 0.014*\"pay\" + 0.013*\"money\" + 0.010*\"work\" + 0.010*\"job\" + 0.009*\"people\" + 0.008*\"year\" + 0.007*\"jobs\" + 0.007*\"tax\" + 0.005*\"item\" + 0.005*\"income\"\n",
      "2017-06-05 21:01:26,649 : INFO : topic #10 (0.067): 0.020*\"code\" + 0.014*\"use\" + 0.007*\"language\" + 0.006*\"using\" + 0.006*\"write\" + 0.005*\"java\" + 0.004*\"memory\" + 0.004*\"systems\" + 0.004*\"run\" + 0.004*\"machine\"\n",
      "2017-06-05 21:01:26,651 : INFO : topic #11 (0.067): 0.006*\"water\" + 0.005*\"energy\" + 0.005*\"power\" + 0.004*\"police\" + 0.004*\"food\" + 0.004*\"city\" + 0.004*\"fiber\" + 0.004*\"air\" + 0.004*\"world\" + 0.003*\"space\"\n",
      "2017-06-05 21:01:26,653 : INFO : topic #7 (0.067): 0.008*\"did\" + 0.007*\"war\" + 0.006*\"history\" + 0.005*\"patent\" + 0.005*\"author\" + 0.005*\"later\" + 0.005*\"art\" + 0.004*\"corp\" + 0.004*\"wasn\" + 0.004*\"read\"\n",
      "2017-06-05 21:01:26,660 : INFO : topic diff=0.028306, rho=0.090536\n",
      "2017-06-05 21:01:26,686 : INFO : PROGRESS: pass 0, at document #246000/246175\n",
      "2017-06-05 21:01:28,173 : INFO : merging changes from 2000 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:28,361 : INFO : topic #8 (0.067): 0.009*\"law\" + 0.008*\"government\" + 0.007*\"right\" + 0.006*\"state\" + 0.006*\"people\" + 0.005*\"trump\" + 0.005*\"power\" + 0.005*\"bad\" + 0.005*\"like\" + 0.005*\"legal\"\n",
      "2017-06-05 21:01:28,363 : INFO : topic #3 (0.067): 0.012*\"use\" + 0.008*\"just\" + 0.007*\"information\" + 0.006*\"user\" + 0.006*\"traffic\" + 0.006*\"does\" + 0.006*\"want\" + 0.005*\"access\" + 0.005*\"phone\" + 0.005*\"using\"\n",
      "2017-06-05 21:01:28,366 : INFO : topic #0 (0.067): 0.009*\"uber\" + 0.008*\"company\" + 0.008*\"market\" + 0.008*\"companies\" + 0.008*\"cost\" + 0.008*\"business\" + 0.006*\"price\" + 0.006*\"car\" + 0.006*\"people\" + 0.005*\"long\"\n",
      "2017-06-05 21:01:28,368 : INFO : topic #13 (0.067): 0.015*\"time\" + 0.014*\"problem\" + 0.012*\"good\" + 0.009*\"ve\" + 0.009*\"game\" + 0.008*\"things\" + 0.007*\"just\" + 0.007*\"problems\" + 0.007*\"lot\" + 0.006*\"better\"\n",
      "2017-06-05 21:01:28,370 : INFO : topic #11 (0.067): 0.006*\"water\" + 0.005*\"light\" + 0.005*\"energy\" + 0.005*\"power\" + 0.004*\"food\" + 0.004*\"police\" + 0.004*\"city\" + 0.004*\"fiber\" + 0.004*\"air\" + 0.004*\"world\"\n",
      "2017-06-05 21:01:28,377 : INFO : topic diff=0.032446, rho=0.090167\n",
      "2017-06-05 21:01:28,846 : INFO : -9.446 per-word bound, 697.5 perplexity estimate based on a held-out corpus of 175 documents with 6007 words\n",
      "2017-06-05 21:01:28,846 : INFO : PROGRESS: pass 0, at document #246175/246175\n",
      "2017-06-05 21:01:28,985 : INFO : merging changes from 175 documents into a model of 246175 documents\n",
      "2017-06-05 21:01:29,173 : INFO : topic #14 (0.067): 0.023*\"data\" + 0.017*\"like\" + 0.010*\"use\" + 0.008*\"type\" + 0.007*\"case\" + 0.007*\"used\" + 0.007*\"read\" + 0.006*\"looks\" + 0.006*\"types\" + 0.005*\"example\"\n",
      "2017-06-05 21:01:29,176 : INFO : topic #6 (0.067): 0.021*\"don\" + 0.018*\"just\" + 0.018*\"think\" + 0.017*\"like\" + 0.015*\"really\" + 0.012*\"people\" + 0.010*\"work\" + 0.009*\"need\" + 0.009*\"know\" + 0.008*\"way\"\n",
      "2017-06-05 21:01:29,178 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.007*\"like\" + 0.006*\"hn\" + 0.006*\"people\" + 0.006*\"ve\" + 0.006*\"social\" + 0.005*\"title\" + 0.004*\"friends\" + 0.004*\"didn\"\n",
      "2017-06-05 21:01:29,180 : INFO : topic #1 (0.067): 0.025*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.008*\"doesn\" + 0.008*\"say\" + 0.007*\"mean\" + 0.007*\"just\" + 0.007*\"point\" + 0.007*\"know\" + 0.007*\"does\"\n",
      "2017-06-05 21:01:29,182 : INFO : topic #8 (0.067): 0.009*\"government\" + 0.009*\"law\" + 0.007*\"right\" + 0.006*\"state\" + 0.006*\"power\" + 0.005*\"people\" + 0.005*\"trump\" + 0.005*\"like\" + 0.005*\"bad\" + 0.004*\"public\"\n",
      "2017-06-05 21:01:29,188 : INFO : topic diff=0.033560, rho=0.089803\n"
     ]
    }
   ],
   "source": [
    "# Create lda model (equivalent to \"fit\" in sklearn)\n",
    "# passes is the max value to keep on iterating until we converge\n",
    "lda = models.LdaModel(corpus=corpus, num_topics=15, id2word=id2word, passes=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's see the 100 most important words for each of the topics. Below, we see the top probabilities of words that would appear in this topic. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2017-06-05 21:01:29,212 : INFO : topic #0 (0.067): 0.008*\"companies\" + 0.008*\"uber\" + 0.008*\"cost\" + 0.008*\"company\" + 0.008*\"market\" + 0.008*\"business\" + 0.006*\"price\" + 0.006*\"people\" + 0.006*\"car\" + 0.005*\"long\" + 0.004*\"money\" + 0.004*\"tech\" + 0.004*\"tip\" + 0.004*\"high\" + 0.004*\"just\" + 0.004*\"self\" + 0.004*\"going\" + 0.004*\"buy\" + 0.003*\"term\" + 0.003*\"costs\" + 0.003*\"quality\" + 0.003*\"point\" + 0.003*\"small\" + 0.003*\"drivers\" + 0.003*\"probably\" + 0.003*\"big\" + 0.003*\"make\" + 0.003*\"model\" + 0.003*\"driving\" + 0.003*\"low\" + 0.003*\"industry\" + 0.003*\"cars\" + 0.003*\"tesla\" + 0.003*\"expensive\" + 0.003*\"time\" + 0.003*\"years\" + 0.003*\"like\" + 0.003*\"ve\" + 0.003*\"new\" + 0.003*\"million\" + 0.003*\"risk\" + 0.002*\"startup\" + 0.002*\"cheap\" + 0.002*\"prices\" + 0.002*\"stock\" + 0.002*\"supply\" + 0.002*\"large\" + 0.002*\"sell\" + 0.002*\"customers\" + 0.002*\"service\" + 0.002*\"value\" + 0.002*\"worth\" + 0.002*\"sales\" + 0.002*\"profit\" + 0.002*\"short\" + 0.002*\"startups\" + 0.002*\"businesses\" + 0.002*\"cheaper\" + 0.002*\"cash\" + 0.002*\"long term\" + 0.002*\"product\" + 0.002*\"lot\" + 0.002*\"meat\" + 0.002*\"good\" + 0.002*\"think\" + 0.002*\"twitter\" + 0.002*\"getting\" + 0.002*\"selling\" + 0.002*\"drive\" + 0.002*\"demand\" + 0.002*\"buying\" + 0.002*\"amazon\" + 0.002*\"making\" + 0.002*\"huge\" + 0.002*\"future\" + 0.002*\"investment\" + 0.002*\"pricing\" + 0.002*\"successful\" + 0.002*\"scale\" + 0.002*\"news\" + 0.002*\"tips\" + 0.002*\"medium\" + 0.002*\"products\" + 0.002*\"technology\" + 0.002*\"success\" + 0.001*\"competition\" + 0.001*\"millions\" + 0.001*\"status\" + 0.001*\"thing\" + 0.001*\"investors\" + 0.001*\"actually\" + 0.001*\"great\" + 0.001*\"growth\" + 0.001*\"hacker\" + 0.001*\"start\" + 0.001*\"dollars\" + 0.001*\"year\" + 0.001*\"right\" + 0.001*\"pretty\" + 0.001*\"day\"\n",
      "2017-06-05 21:01:29,214 : INFO : topic #1 (0.067): 0.025*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.008*\"doesn\" + 0.008*\"say\" + 0.007*\"mean\" + 0.007*\"just\" + 0.007*\"point\" + 0.007*\"know\" + 0.007*\"does\" + 0.006*\"like\" + 0.005*\"question\" + 0.005*\"article\" + 0.005*\"person\" + 0.005*\"make\" + 0.005*\"understand\" + 0.005*\"actually\" + 0.004*\"way\" + 0.004*\"true\" + 0.004*\"fact\" + 0.004*\"wrong\" + 0.004*\"saying\" + 0.004*\"said\" + 0.004*\"believe\" + 0.004*\"evidence\" + 0.004*\"argument\" + 0.004*\"isn\" + 0.004*\"case\" + 0.003*\"human\" + 0.003*\"real\" + 0.003*\"didn\" + 0.003*\"different\" + 0.003*\"comment\" + 0.003*\"men\" + 0.003*\"means\" + 0.003*\"right\" + 0.003*\"reason\" + 0.003*\"talking\" + 0.003*\"sure\" + 0.003*\"did\" + 0.002*\"care\" + 0.002*\"things\" + 0.002*\"science\" + 0.002*\"clear\" + 0.002*\"number\" + 0.002*\"women\" + 0.002*\"answer\" + 0.002*\"really\" + 0.002*\"read\" + 0.002*\"matter\" + 0.002*\"example\" + 0.002*\"words\" + 0.002*\"given\" + 0.002*\"non\" + 0.002*\"value\" + 0.002*\"yes\" + 0.002*\"simply\" + 0.002*\"claim\" + 0.002*\"says\" + 0.002*\"credit\" + 0.002*\"reality\" + 0.002*\"good\" + 0.002*\"based\" + 0.002*\"word\" + 0.002*\"humans\" + 0.002*\"black\" + 0.002*\"aren\" + 0.002*\"agree\" + 0.002*\"exist\" + 0.002*\"change\" + 0.002*\"having\" + 0.002*\"definition\" + 0.002*\"response\" + 0.002*\"possible\" + 0.002*\"explain\" + 0.002*\"view\" + 0.002*\"mb\" + 0.002*\"statement\" + 0.002*\"argue\" + 0.002*\"correct\" + 0.002*\"don know\" + 0.002*\"study\" + 0.002*\"particular\" + 0.002*\"difference\" + 0.002*\"clearly\" + 0.002*\"discussion\" + 0.002*\"meaning\" + 0.002*\"better\" + 0.002*\"doesn mean\" + 0.002*\"general\" + 0.001*\"context\" + 0.001*\"reasonable\" + 0.001*\"life\" + 0.001*\"thought\" + 0.001*\"man\" + 0.001*\"truth\" + 0.001*\"obvious\" + 0.001*\"theory\" + 0.001*\"don think\" + 0.001*\"don understand\"\n",
      "2017-06-05 21:01:29,217 : INFO : topic #2 (0.067): 0.020*\"google\" + 0.012*\"open\" + 0.010*\"content\" + 0.009*\"free\" + 0.009*\"source\" + 0.009*\"don\" + 0.009*\"software\" + 0.008*\"product\" + 0.008*\"users\" + 0.008*\"make\" + 0.007*\"internet\" + 0.007*\"use\" + 0.007*\"facebook\" + 0.006*\"service\" + 0.006*\"want\" + 0.006*\"open source\" + 0.006*\"ads\" + 0.006*\"site\" + 0.006*\"just\" + 0.005*\"like\" + 0.005*\"security\" + 0.005*\"company\" + 0.005*\"network\" + 0.004*\"ad\" + 0.004*\"license\" + 0.004*\"sites\" + 0.004*\"email\" + 0.004*\"money\" + 0.004*\"need\" + 0.004*\"access\" + 0.004*\"business\" + 0.003*\"time\" + 0.003*\"nintendo\" + 0.003*\"app\" + 0.003*\"companies\" + 0.003*\"online\" + 0.003*\"devices\" + 0.003*\"phone\" + 0.003*\"gpl\" + 0.003*\"web\" + 0.003*\"services\" + 0.003*\"doesn\" + 0.003*\"advertising\" + 0.003*\"good\" + 0.003*\"project\" + 0.003*\"know\" + 0.003*\"device\" + 0.003*\"pay\" + 0.002*\"products\" + 0.002*\"user\" + 0.002*\"sell\" + 0.002*\"lock\" + 0.002*\"new\" + 0.002*\"let\" + 0.002*\"llc\" + 0.002*\"idea\" + 0.002*\"support\" + 0.002*\"marketing\" + 0.002*\"sure\" + 0.002*\"people\" + 0.002*\"networks\" + 0.002*\"account\" + 0.002*\"big\" + 0.002*\"angular\" + 0.002*\"ll\" + 0.002*\"website\" + 0.002*\"isn\" + 0.002*\"projects\" + 0.002*\"provide\" + 0.002*\"paid\" + 0.002*\"available\" + 0.002*\"microsoft\" + 0.002*\"did\" + 0.002*\"freebsd\" + 0.002*\"commercial\" + 0.002*\"bitcoin\" + 0.002*\"making\" + 0.002*\"won\" + 0.002*\"fake\" + 0.002*\"digital\" + 0.002*\"work\" + 0.002*\"tracking\" + 0.002*\"upgrade\" + 0.002*\"slack\" + 0.002*\"going\" + 0.002*\"community\" + 0.002*\"using\" + 0.002*\"search\" + 0.001*\"personal\" + 0.001*\"way\" + 0.001*\"founder\" + 0.001*\"websites\" + 0.001*\"offer\" + 0.001*\"non\" + 0.001*\"block\" + 0.001*\"messages\" + 0.001*\"smartphone\" + 0.001*\"customers\" + 0.001*\"drm\" + 0.001*\"create\"\n",
      "2017-06-05 21:01:29,219 : INFO : topic #3 (0.067): 0.013*\"use\" + 0.008*\"just\" + 0.007*\"user\" + 0.007*\"information\" + 0.006*\"does\" + 0.006*\"want\" + 0.006*\"traffic\" + 0.005*\"android\" + 0.005*\"phone\" + 0.005*\"using\" + 0.005*\"run\" + 0.005*\"pulseaudio\" + 0.004*\"access\" + 0.004*\"audio\" + 0.004*\"service\" + 0.004*\"need\" + 0.004*\"don\" + 0.004*\"ve\" + 0.004*\"share\" + 0.004*\"ip\" + 0.004*\"alsa\" + 0.004*\"data\" + 0.004*\"linux\" + 0.003*\"account\" + 0.003*\"desktop\" + 0.003*\"kernel\" + 0.003*\"different\" + 0.003*\"click\" + 0.003*\"doesn\" + 0.003*\"apng\" + 0.003*\"sure\" + 0.003*\"services\" + 0.003*\"like\" + 0.003*\"format\" + 0.003*\"key\" + 0.003*\"work\" + 0.003*\"ios\" + 0.003*\"address\" + 0.003*\"home\" + 0.003*\"case\" + 0.003*\"number\" + 0.002*\"having\" + 0.002*\"switch\" + 0.002*\"running\" + 0.002*\"netflix\" + 0.002*\"cloud\" + 0.002*\"party\" + 0.002*\"email\" + 0.002*\"bike\" + 0.002*\"tried\" + 0.002*\"trust\" + 0.002*\"aws\" + 0.002*\"password\" + 0.002*\"multiple\" + 0.002*\"signal\" + 0.002*\"app\" + 0.002*\"gif\" + 0.002*\"tv\" + 0.002*\"host\" + 0.002*\"connection\" + 0.002*\"way\" + 0.002*\"completely\" + 0.002*\"freedom\" + 0.002*\"able\" + 0.002*\"network\" + 0.002*\"cross\" + 0.002*\"info\" + 0.002*\"wouldn\" + 0.002*\"sent\" + 0.002*\"request\" + 0.002*\"scroll\" + 0.002*\"isp\" + 0.002*\"used\" + 0.002*\"mobile\" + 0.002*\"video\" + 0.002*\"voice\" + 0.002*\"time\" + 0.002*\"provider\" + 0.002*\"isps\" + 0.002*\"public\" + 0.002*\"single\" + 0.002*\"platform\" + 0.002*\"docker\" + 0.002*\"mail\" + 0.002*\"sign\" + 0.002*\"hosting\" + 0.002*\"make\" + 0.002*\"fb\" + 0.002*\"player\" + 0.002*\"set\" + 0.002*\"debian\" + 0.002*\"computer\" + 0.002*\"send\" + 0.002*\"hardware\" + 0.002*\"control\" + 0.002*\"music\" + 0.002*\"know\" + 0.002*\"store\" + 0.001*\"accounts\" + 0.001*\"lot\"\n",
      "2017-06-05 21:01:29,221 : INFO : topic #4 (0.067): 0.011*\"life\" + 0.008*\"post\" + 0.007*\"like\" + 0.006*\"hn\" + 0.006*\"people\" + 0.006*\"ve\" + 0.006*\"social\" + 0.005*\"title\" + 0.004*\"friends\" + 0.004*\"didn\" + 0.004*\"got\" + 0.004*\"thing\" + 0.004*\"search\" + 0.004*\"comment\" + 0.004*\"point\" + 0.004*\"just\" + 0.004*\"pro\" + 0.003*\"read\" + 0.003*\"commit\" + 0.003*\"happy\" + 0.003*\"link\" + 0.003*\"right\" + 0.003*\"think\" + 0.003*\"did\" + 0.003*\"movie\" + 0.003*\"story\" + 0.003*\"way\" + 0.003*\"thanks\" + 0.003*\"privacy\" + 0.003*\"oh\" + 0.003*\"battery\" + 0.003*\"kids\" + 0.003*\"laptop\" + 0.003*\"family\" + 0.003*\"interesting\" + 0.003*\"parent\" + 0.003*\"maybe\" + 0.002*\"ratings\" + 0.002*\"thought\" + 0.002*\"comments\" + 0.002*\"new\" + 0.002*\"away\" + 0.002*\"film\" + 0.002*\"media\" + 0.002*\"yes\" + 0.002*\"thread\" + 0.002*\"probably\" + 0.002*\"stories\" + 0.002*\"article\" + 0.002*\"ask\" + 0.002*\"bsd\" + 0.002*\"seen\" + 0.002*\"treatment\" + 0.002*\"audience\" + 0.002*\"kind\" + 0.002*\"guess\" + 0.002*\"don\" + 0.002*\"stop\" + 0.002*\"community\" + 0.002*\"child\" + 0.002*\"hope\" + 0.002*\"ll\" + 0.002*\"blog post\" + 0.002*\"star\" + 0.002*\"said\" + 0.002*\"time\" + 0.002*\"person\" + 0.002*\"ah\" + 0.002*\"movies\" + 0.002*\"aware\" + 0.002*\"wasn\" + 0.002*\"know\" + 0.002*\"good\" + 0.002*\"old\" + 0.002*\"original\" + 0.002*\"day\" + 0.002*\"defend\" + 0.002*\"networking\" + 0.001*\"sure\" + 0.001*\"page\" + 0.001*\"cells\" + 0.001*\"mention\" + 0.001*\"calling\" + 0.001*\"looking\" + 0.001*\"arrow\" + 0.001*\"care\" + 0.001*\"agents\" + 0.001*\"children\" + 0.001*\"sorry\" + 0.001*\"interaction\" + 0.001*\"agent\" + 0.001*\"edit\" + 0.001*\"seeing\" + 0.001*\"self\" + 0.001*\"proof\" + 0.001*\"editing\" + 0.001*\"reading\" + 0.001*\"links\" + 0.001*\"phrase\" + 0.001*\"taken\"\n",
      "2017-06-05 21:01:29,224 : INFO : topic #5 (0.067): 0.023*\"https\" + 0.016*\"org\" + 0.013*\"github\" + 0.011*\"windows\" + 0.011*\"wikipedia\" + 0.011*\"web\" + 0.009*\"en\" + 0.009*\"wiki\" + 0.009*\"com\" + 0.009*\"use\" + 0.008*\"github com\" + 0.008*\"org wiki\" + 0.008*\"wikipedia org\" + 0.008*\"https github\" + 0.008*\"https en\" + 0.008*\"en wikipedia\" + 0.007*\"browser\" + 0.007*\"linux\" + 0.007*\"using\" + 0.007*\"page\" + 0.006*\"server\" + 0.006*\"support\" + 0.006*\"file\" + 0.006*\"apps\" + 0.005*\"react\" + 0.005*\"app\" + 0.005*\"html\" + 0.005*\"gitlab\" + 0.005*\"version\" + 0.004*\"os\" + 0.004*\"text\" + 0.004*\"install\" + 0.004*\"just\" + 0.004*\"like\" + 0.004*\"issues\" + 0.004*\"microsoft\" + 0.004*\"files\" + 0.004*\"chrome\" + 0.004*\"firefox\" + 0.003*\"git\" + 0.003*\"io\" + 0.003*\"used\" + 0.003*\"http\" + 0.003*\"user\" + 0.003*\"ms\" + 0.003*\"running\" + 0.003*\"css\" + 0.003*\"native\" + 0.003*\"ui\" + 0.003*\"open\" + 0.003*\"release\" + 0.003*\"client\" + 0.003*\"download\" + 0.003*\"docs\" + 0.002*\"works\" + 0.002*\"mac\" + 0.002*\"default\" + 0.002*\"update\" + 0.002*\"browsers\" + 0.002*\"source\" + 0.002*\"updates\" + 0.002*\"master\" + 0.002*\"json\" + 0.002*\"run\" + 0.002*\"window\" + 0.002*\"does\" + 0.002*\"work\" + 0.002*\"add\" + 0.002*\"aframe\" + 0.002*\"bar\" + 0.002*\"jpeg\" + 0.002*\"api\" + 0.002*\"supported\" + 0.002*\"new\" + 0.002*\"nice\" + 0.002*\"users\" + 0.002*\"module\" + 0.002*\"command\" + 0.002*\"edit\" + 0.002*\"display\" + 0.002*\"site\" + 0.002*\"screen\" + 0.002*\"list\" + 0.002*\"stack\" + 0.002*\"unix\" + 0.002*\"pages\" + 0.002*\"keys\" + 0.002*\"link\" + 0.002*\"emacs\" + 0.002*\"web components\" + 0.002*\"blob\" + 0.002*\"keyboard\" + 0.001*\"based\" + 0.001*\"archive\" + 0.001*\"terminal\" + 0.001*\"released\" + 0.001*\"mobile\" + 0.001*\"example\" + 0.001*\"delaware\" + 0.001*\"roundup\"\n",
      "2017-06-05 21:01:29,227 : INFO : topic #6 (0.067): 0.021*\"don\" + 0.018*\"just\" + 0.018*\"think\" + 0.017*\"like\" + 0.015*\"really\" + 0.012*\"people\" + 0.010*\"work\" + 0.009*\"need\" + 0.009*\"know\" + 0.008*\"way\" + 0.008*\"time\" + 0.008*\"want\" + 0.007*\"make\" + 0.007*\"ve\" + 0.007*\"things\" + 0.006*\"lot\" + 0.005*\"good\" + 0.005*\"feel\" + 0.005*\"makes\" + 0.005*\"better\" + 0.005*\"doing\" + 0.005*\"hard\" + 0.004*\"thing\" + 0.004*\"going\" + 0.004*\"sense\" + 0.004*\"learn\" + 0.004*\"experience\" + 0.004*\"ll\" + 0.003*\"programming\" + 0.003*\"don think\" + 0.003*\"actually\" + 0.003*\"stuff\" + 0.003*\"probably\" + 0.003*\"years\" + 0.003*\"end\" + 0.003*\"understand\" + 0.003*\"day\" + 0.003*\"don know\" + 0.003*\"new\" + 0.003*\"learning\" + 0.003*\"right\" + 0.003*\"working\" + 0.002*\"having\" + 0.002*\"point\" + 0.002*\"math\" + 0.002*\"bit\" + 0.002*\"course\" + 0.002*\"little\" + 0.002*\"language\" + 0.002*\"job\" + 0.002*\"look\" + 0.002*\"kind\" + 0.002*\"doesn\" + 0.002*\"trying\" + 0.002*\"best\" + 0.002*\"maybe\" + 0.002*\"come\" + 0.002*\"start\" + 0.002*\"level\" + 0.002*\"help\" + 0.002*\"getting\" + 0.002*\"say\" + 0.002*\"great\" + 0.002*\"sure\" + 0.002*\"different\" + 0.002*\"scala\" + 0.002*\"able\" + 0.002*\"pretty\" + 0.002*\"try\" + 0.002*\"idea\" + 0.002*\"isn\" + 0.002*\"looking\" + 0.002*\"feel like\" + 0.002*\"skills\" + 0.002*\"book\" + 0.002*\"agree\" + 0.002*\"easy\" + 0.002*\"started\" + 0.001*\"worked\" + 0.001*\"makes sense\" + 0.001*\"knowledge\" + 0.001*\"important\" + 0.001*\"components\" + 0.001*\"love\" + 0.001*\"easier\" + 0.001*\"thinking\" + 0.001*\"simple\" + 0.001*\"mind\" + 0.001*\"got\" + 0.001*\"quite\" + 0.001*\"especially\" + 0.001*\"advice\" + 0.001*\"school\" + 0.001*\"world\" + 0.001*\"common\" + 0.001*\"problem\" + 0.001*\"making\" + 0.001*\"ideas\" + 0.001*\"don want\" + 0.001*\"instead\"\n",
      "2017-06-05 21:01:29,229 : INFO : topic #7 (0.067): 0.008*\"women\" + 0.008*\"did\" + 0.007*\"war\" + 0.006*\"history\" + 0.005*\"patent\" + 0.004*\"later\" + 0.004*\"author\" + 0.004*\"read\" + 0.004*\"years\" + 0.004*\"wasn\" + 0.004*\"art\" + 0.004*\"corp\" + 0.004*\"man\" + 0.004*\"say\" + 0.003*\"remember\" + 0.003*\"russia\" + 0.003*\"female\" + 0.003*\"japanese\" + 0.003*\"patents\" + 0.003*\"culture\" + 0.003*\"lyft\" + 0.003*\"didn\" + 0.003*\"world\" + 0.003*\"reading\" + 0.003*\"just\" + 0.003*\"said\" + 0.003*\"comments\" + 0.003*\"china\" + 0.003*\"guy\" + 0.002*\"came\" + 0.002*\"english\" + 0.002*\"god\" + 0.002*\"reddit\" + 0.002*\"male\" + 0.002*\"th\" + 0.002*\"notes\" + 0.002*\"yep\" + 0.002*\"russian\" + 0.002*\"today\" + 0.002*\"comment\" + 0.002*\"dr\" + 0.002*\"recall\" + 0.002*\"totally\" + 0.002*\"likely\" + 0.002*\"usd\" + 0.002*\"restaurant\" + 0.002*\"chinese\" + 0.002*\"remain\" + 0.002*\"wonder\" + 0.002*\"time\" + 0.002*\"gun\" + 0.002*\"original\" + 0.002*\"like\" + 0.002*\"woman\" + 0.002*\"late\" + 0.002*\"fuck\" + 0.002*\"story\" + 0.002*\"early\" + 0.002*\"coffee\" + 0.002*\"mit\" + 0.002*\"thought\" + 0.002*\"right\" + 0.002*\"got\" + 0.002*\"years later\" + 0.002*\"bias\" + 0.002*\"sum\" + 0.002*\"databases\" + 0.002*\"labour\" + 0.002*\"restrictions\" + 0.002*\"gold\" + 0.001*\"codecs\" + 0.001*\"macos\" + 0.001*\"directed\" + 0.001*\"neat\" + 0.001*\"entertainment\" + 0.001*\"section\" + 0.001*\"analytics\" + 0.001*\"origin\" + 0.001*\"going\" + 0.001*\"tl\" + 0.001*\"paper\" + 0.001*\"long\" + 0.001*\"imbalance\" + 0.001*\"internet\" + 0.001*\"onions\" + 0.001*\"agreed\" + 0.001*\"tl dr\" + 0.001*\"sticker\" + 0.001*\"honest\" + 0.001*\"headline\" + 0.001*\"mind\" + 0.001*\"pre\" + 0.001*\"century\" + 0.001*\"ass\" + 0.001*\"artificial\" + 0.001*\"referring\" + 0.001*\"sex\" + 0.001*\"far\" + 0.001*\"shock\" + 0.001*\"true\"\n",
      "2017-06-05 21:01:29,232 : INFO : topic #8 (0.067): 0.009*\"government\" + 0.009*\"law\" + 0.007*\"right\" + 0.006*\"state\" + 0.006*\"power\" + 0.005*\"people\" + 0.005*\"trump\" + 0.005*\"like\" + 0.005*\"bad\" + 0.004*\"public\" + 0.004*\"legal\" + 0.004*\"states\" + 0.004*\"political\" + 0.004*\"society\" + 0.003*\"laws\" + 0.003*\"age\" + 0.003*\"court\" + 0.003*\"policy\" + 0.003*\"think\" + 0.003*\"world\" + 0.003*\"way\" + 0.003*\"risk\" + 0.003*\"vote\" + 0.002*\"party\" + 0.002*\"dating\" + 0.002*\"rights\" + 0.002*\"free\" + 0.002*\"anti\" + 0.002*\"rules\" + 0.002*\"want\" + 0.002*\"thing\" + 0.002*\"social\" + 0.002*\"just\" + 0.002*\"best\" + 0.002*\"interests\" + 0.002*\"act\" + 0.002*\"good\" + 0.002*\"current\" + 0.002*\"change\" + 0.002*\"idea\" + 0.002*\"isn\" + 0.002*\"regulation\" + 0.002*\"case\" + 0.002*\"place\" + 0.002*\"united\" + 0.002*\"majority\" + 0.002*\"american\" + 0.002*\"economic\" + 0.002*\"win\" + 0.002*\"views\" + 0.002*\"moral\" + 0.002*\"kind\" + 0.002*\"likely\" + 0.002*\"possible\" + 0.002*\"voting\" + 0.002*\"regulations\" + 0.002*\"does\" + 0.002*\"private\" + 0.002*\"result\" + 0.002*\"wing\" + 0.002*\"human\" + 0.002*\"united states\" + 0.002*\"control\" + 0.001*\"let\" + 0.001*\"decision\" + 0.001*\"agree\" + 0.001*\"effect\" + 0.001*\"federal\" + 0.001*\"left\" + 0.001*\"san\" + 0.001*\"order\" + 0.001*\"means\" + 0.001*\"citizens\" + 0.001*\"issue\" + 0.001*\"net\" + 0.001*\"position\" + 0.001*\"making\" + 0.001*\"certain\" + 0.001*\"era\" + 0.001*\"fight\" + 0.001*\"rule\" + 0.001*\"individual\" + 0.001*\"example\" + 0.001*\"important\" + 0.001*\"policies\" + 0.001*\"property\" + 0.001*\"fly\" + 0.001*\"orders\" + 0.001*\"chance\" + 0.001*\"defense\" + 0.001*\"better\" + 0.001*\"market\" + 0.001*\"action\" + 0.001*\"support\" + 0.001*\"wouldn\" + 0.001*\"research\" + 0.001*\"democracy\" + 0.001*\"financial\" + 0.001*\"given\" + 0.001*\"culture\"\n",
      "2017-06-05 21:01:29,234 : INFO : topic #9 (0.067): 0.014*\"pay\" + 0.013*\"money\" + 0.010*\"work\" + 0.009*\"job\" + 0.009*\"people\" + 0.008*\"year\" + 0.007*\"jobs\" + 0.007*\"tax\" + 0.005*\"item\" + 0.005*\"rate\" + 0.005*\"paying\" + 0.005*\"home\" + 0.005*\"income\" + 0.005*\"ycombinator\" + 0.005*\"paid\" + 0.005*\"company\" + 0.005*\"country\" + 0.005*\"ycombinator com\" + 0.005*\"id\" + 0.005*\"https news\" + 0.005*\"living\" + 0.005*\"news ycombinator\" + 0.005*\"news\" + 0.004*\"time\" + 0.004*\"taxes\" + 0.004*\"hours\" + 0.004*\"item id\" + 0.004*\"com item\" + 0.004*\"live\" + 0.004*\"tipping\" + 0.004*\"debt\" + 0.004*\"don\" + 0.004*\"countries\" + 0.003*\"working\" + 0.003*\"driver\" + 0.003*\"health\" + 0.003*\"higher\" + 0.003*\"high\" + 0.003*\"insurance\" + 0.003*\"school\" + 0.003*\"government\" + 0.003*\"cost\" + 0.003*\"house\" + 0.003*\"poor\" + 0.003*\"education\" + 0.003*\"lower\" + 0.003*\"city\" + 0.003*\"month\" + 0.003*\"workers\" + 0.003*\"employees\" + 0.003*\"uk\" + 0.002*\"healthcare\" + 0.002*\"salary\" + 0.002*\"companies\" + 0.002*\"rates\" + 0.002*\"employee\" + 0.002*\"wage\" + 0.002*\"labor\" + 0.002*\"years\" + 0.002*\"rural\" + 0.002*\"college\" + 0.002*\"day\" + 0.002*\"class\" + 0.002*\"rating\" + 0.002*\"office\" + 0.002*\"employer\" + 0.002*\"average\" + 0.002*\"costs\" + 0.002*\"care\" + 0.002*\"getting\" + 0.002*\"spend\" + 0.002*\"america\" + 0.002*\"minimum\" + 0.002*\"hire\" + 0.002*\"family\" + 0.002*\"rent\" + 0.002*\"week\" + 0.002*\"lot\" + 0.002*\"kids\" + 0.002*\"local\" + 0.002*\"rich\" + 0.002*\"leave\" + 0.002*\"just\" + 0.002*\"low\" + 0.002*\"afford\" + 0.002*\"life\" + 0.002*\"area\" + 0.002*\"housing\" + 0.002*\"students\" + 0.002*\"capital\" + 0.002*\"places\" + 0.002*\"schools\" + 0.002*\"american\" + 0.002*\"million\" + 0.002*\"public\" + 0.002*\"wages\" + 0.002*\"middle\" + 0.002*\"remote\" + 0.002*\"free\" + 0.002*\"spending\"\n",
      "2017-06-05 21:01:29,237 : INFO : topic #10 (0.067): 0.021*\"code\" + 0.013*\"use\" + 0.007*\"java\" + 0.007*\"language\" + 0.006*\"using\" + 0.006*\"write\" + 0.004*\"memory\" + 0.004*\"systems\" + 0.004*\"time\" + 0.004*\"used\" + 0.004*\"like\" + 0.004*\"machine\" + 0.004*\"run\" + 0.003*\"build\" + 0.003*\"js\" + 0.003*\"need\" + 0.003*\"performance\" + 0.003*\"simple\" + 0.003*\"project\" + 0.003*\"library\" + 0.003*\"rust\" + 0.003*\"python\" + 0.003*\"design\" + 0.003*\"just\" + 0.003*\"easy\" + 0.003*\"tools\" + 0.003*\"features\" + 0.003*\"level\" + 0.003*\"does\" + 0.003*\"javascript\" + 0.003*\"writing\" + 0.003*\"written\" + 0.003*\"learning\" + 0.003*\"way\" + 0.003*\"api\" + 0.002*\"compiler\" + 0.002*\"program\" + 0.002*\"languages\" + 0.002*\"developers\" + 0.002*\"software\" + 0.002*\"really\" + 0.002*\"implementation\" + 0.002*\"new\" + 0.002*\"standard\" + 0.002*\"running\" + 0.002*\"development\" + 0.002*\"based\" + 0.002*\"lot\" + 0.002*\"example\" + 0.002*\"different\" + 0.002*\"hardware\" + 0.002*\"built\" + 0.002*\"support\" + 0.002*\"image\" + 0.002*\"application\" + 0.002*\"want\" + 0.002*\"possible\" + 0.002*\"great\" + 0.002*\"small\" + 0.002*\"libraries\" + 0.002*\"fast\" + 0.002*\"source\" + 0.002*\"faster\" + 0.002*\"better\" + 0.002*\"framework\" + 0.002*\"pretty\" + 0.002*\"easier\" + 0.002*\"vs\" + 0.002*\"line\" + 0.002*\"nice\" + 0.002*\"lines\" + 0.002*\"actually\" + 0.002*\"feature\" + 0.002*\"applications\" + 0.002*\"make\" + 0.002*\"set\" + 0.002*\"size\" + 0.002*\"stuff\" + 0.002*\"uses\" + 0.002*\"useful\" + 0.002*\"interesting\" + 0.002*\"dom\" + 0.002*\"core\" + 0.002*\"cpu\" + 0.002*\"test\" + 0.002*\"architecture\" + 0.002*\"single\" + 0.002*\"designed\" + 0.002*\"complex\" + 0.002*\"able\" + 0.002*\"end\" + 0.002*\"speed\" + 0.002*\"haskell\" + 0.002*\"implement\" + 0.002*\"compatible\" + 0.002*\"static\" + 0.002*\"source code\" + 0.002*\"instead\" + 0.002*\"model\" + 0.002*\"load\"\n",
      "2017-06-05 21:01:29,239 : INFO : topic #11 (0.067): 0.006*\"water\" + 0.005*\"light\" + 0.005*\"energy\" + 0.005*\"power\" + 0.004*\"police\" + 0.004*\"food\" + 0.004*\"city\" + 0.003*\"world\" + 0.003*\"fiber\" + 0.003*\"space\" + 0.003*\"air\" + 0.003*\"monsanto\" + 0.003*\"papers\" + 0.003*\"mass\" + 0.003*\"cold\" + 0.003*\"red\" + 0.003*\"new\" + 0.003*\"large\" + 0.002*\"like\" + 0.002*\"hole\" + 0.002*\"berkeley\" + 0.002*\"area\" + 0.002*\"pdf\" + 0.002*\"far\" + 0.002*\"earth\" + 0.002*\"global\" + 0.002*\"high\" + 0.002*\"drug\" + 0.002*\"times\" + 0.002*\"humanity\" + 0.002*\"military\" + 0.002*\"live\" + 0.002*\"article\" + 0.002*\"flight\" + 0.002*\"pressure\" + 0.002*\"nuclear\" + 0.002*\"body\" + 0.002*\"glyphosate\" + 0.002*\"population\" + 0.002*\"crime\" + 0.002*\"safe\" + 0.002*\"black\" + 0.002*\"small\" + 0.002*\"paper\" + 0.002*\"weight\" + 0.002*\"eat\" + 0.002*\"just\" + 0.002*\"land\" + 0.002*\"weapons\" + 0.002*\"budget\" + 0.002*\"cities\" + 0.002*\"australia\" + 0.002*\"effects\" + 0.002*\"cause\" + 0.002*\"black hole\" + 0.002*\"areas\" + 0.002*\"dangerous\" + 0.002*\"climate\" + 0.002*\"heat\" + 0.002*\"research\" + 0.002*\"speed\" + 0.002*\"scale\" + 0.002*\"quite\" + 0.002*\"place\" + 0.002*\"ideal\" + 0.002*\"near\" + 0.002*\"south\" + 0.002*\"aircraft\" + 0.002*\"comma\" + 0.002*\"case\" + 0.001*\"fear\" + 0.001*\"away\" + 0.001*\"human\" + 0.001*\"low\" + 0.001*\"lot\" + 0.001*\"german\" + 0.001*\"normal\" + 0.001*\"ground\" + 0.001*\"temperature\" + 0.001*\"north\" + 0.001*\"possible\" + 0.001*\"york\" + 0.001*\"inside\" + 0.001*\"feet\" + 0.001*\"new york\" + 0.001*\"offline\" + 0.001*\"kill\" + 0.001*\"plane\" + 0.001*\"close\" + 0.001*\"likely\" + 0.001*\"obama\" + 0.001*\"pretty\" + 0.001*\"traffic light\" + 0.001*\"probably\" + 0.001*\"french\" + 0.001*\"big\" + 0.001*\"safety\" + 0.001*\"produce\" + 0.001*\"glass\" + 0.001*\"bit\"\n",
      "2017-06-05 21:01:29,241 : INFO : topic #12 (0.067): 0.052*\"com\" + 0.038*\"https\" + 0.033*\"www\" + 0.029*\"http\" + 0.018*\"https www\" + 0.014*\"http www\" + 0.012*\"ago\" + 0.011*\"video\" + 0.011*\"years\" + 0.009*\"youtube\" + 0.009*\"news\" + 0.008*\"watch\" + 0.007*\"google\" + 0.007*\"like\" + 0.007*\"blog\" + 0.006*\"years ago\" + 0.006*\"org\" + 0.005*\"videos\" + 0.004*\"apple\" + 0.004*\"kotlin\" + 0.004*\"article\" + 0.004*\"amazon\" + 0.004*\"great\" + 0.004*\"html\" + 0.003*\"book\" + 0.003*\"youtube com\" + 0.003*\"sounds\" + 0.003*\"articles\" + 0.003*\"www youtube\" + 0.003*\"com watch\" + 0.003*\"year\" + 0.003*\"new\" + 0.003*\"day\" + 0.003*\"bay\" + 0.003*\"interesting\" + 0.002*\"yc\" + 0.002*\"google com\" + 0.002*\"net\" + 0.002*\"months\" + 0.002*\"link\" + 0.002*\"sounds like\" + 0.002*\"cloud\" + 0.002*\"aws\" + 0.002*\"looking\" + 0.002*\"gaming\" + 0.002*\"couple\" + 0.002*\"bay area\" + 0.002*\"vim\" + 0.002*\"recommend\" + 0.002*\"camera\" + 0.002*\"site\" + 0.002*\"vr\" + 0.002*\"days\" + 0.002*\"saw\" + 0.002*\"gov\" + 0.002*\"png\" + 0.002*\"page\" + 0.002*\"recently\" + 0.002*\"ipad\" + 0.002*\"time\" + 0.002*\"dropbox\" + 0.002*\"books\" + 0.002*\"edu\" + 0.002*\"ve\" + 0.002*\"archive\" + 0.002*\"launch\" + 0.002*\"bought\" + 0.002*\"week\" + 0.002*\"providers\" + 0.002*\"amazing\" + 0.002*\"good\" + 0.002*\"reading\" + 0.002*\"used\" + 0.002*\"similar\" + 0.002*\"designer\" + 0.002*\"lyrics\" + 0.002*\"jack\" + 0.001*\"drive\" + 0.001*\"search\" + 0.001*\"channel\" + 0.001*\"games\" + 0.001*\"looks\" + 0.001*\"post\" + 0.001*\"linear\" + 0.001*\"remember\" + 0.001*\"topic\" + 0.001*\"recent\" + 0.001*\"uk\" + 0.001*\"look\" + 0.001*\"gb\" + 0.001*\"check\" + 0.001*\"edit\" + 0.001*\"interested\" + 0.001*\"did\" + 0.001*\"pretty\" + 0.001*\"read\" + 0.001*\"story\" + 0.001*\"series\" + 0.001*\"got\" + 0.001*\"best\"\n",
      "2017-06-05 21:01:29,244 : INFO : topic #13 (0.067): 0.015*\"time\" + 0.014*\"problem\" + 0.012*\"good\" + 0.009*\"game\" + 0.009*\"ve\" + 0.008*\"things\" + 0.007*\"just\" + 0.007*\"problems\" + 0.007*\"lot\" + 0.007*\"years\" + 0.006*\"better\" + 0.006*\"games\" + 0.005*\"real\" + 0.005*\"like\" + 0.005*\"old\" + 0.005*\"work\" + 0.005*\"long\" + 0.005*\"thanks\" + 0.004*\"ll\" + 0.004*\"interview\" + 0.004*\"ai\" + 0.004*\"new\" + 0.004*\"sure\" + 0.004*\"questions\" + 0.004*\"maybe\" + 0.003*\"great\" + 0.003*\"solve\" + 0.003*\"software\" + 0.003*\"issues\" + 0.003*\"going\" + 0.003*\"thing\" + 0.003*\"world\" + 0.003*\"know\" + 0.003*\"play\" + 0.003*\"pretty\" + 0.003*\"way\" + 0.003*\"test\" + 0.003*\"year\" + 0.003*\"team\" + 0.003*\"really\" + 0.003*\"whiteboard\" + 0.003*\"actually\" + 0.003*\"tell\" + 0.003*\"bad\" + 0.003*\"issue\" + 0.003*\"solution\" + 0.003*\"did\" + 0.003*\"fix\" + 0.003*\"testing\" + 0.003*\"isn\" + 0.002*\"heard\" + 0.002*\"haven\" + 0.002*\"tests\" + 0.002*\"try\" + 0.002*\"definitely\" + 0.002*\"talk\" + 0.002*\"quality\" + 0.002*\"kind\" + 0.002*\"best\" + 0.002*\"process\" + 0.002*\"big\" + 0.002*\"having\" + 0.002*\"long time\" + 0.002*\"far\" + 0.002*\"got\" + 0.002*\"didn\" + 0.002*\"fun\" + 0.002*\"don\" + 0.002*\"trying\" + 0.002*\"engineering\" + 0.002*\"working\" + 0.002*\"different\" + 0.002*\"think\" + 0.002*\"candidate\" + 0.002*\"right\" + 0.002*\"idea\" + 0.002*\"seen\" + 0.002*\"luck\" + 0.002*\"playing\" + 0.002*\"thank\" + 0.002*\"hard\" + 0.002*\"probably\" + 0.002*\"algorithm\" + 0.002*\"looking\" + 0.002*\"answer\" + 0.002*\"interviews\" + 0.002*\"start\" + 0.002*\"help\" + 0.002*\"people\" + 0.002*\"technical\" + 0.002*\"cool\" + 0.002*\"developers\" + 0.002*\"works\" + 0.002*\"past\" + 0.002*\"ask\" + 0.001*\"mario\" + 0.001*\"doing\" + 0.001*\"solved\" + 0.001*\"change\" + 0.001*\"want\"\n",
      "2017-06-05 21:01:29,246 : INFO : topic #14 (0.067): 0.023*\"data\" + 0.017*\"like\" + 0.010*\"use\" + 0.008*\"type\" + 0.007*\"case\" + 0.007*\"used\" + 0.007*\"read\" + 0.006*\"looks\" + 0.006*\"types\" + 0.005*\"example\" + 0.005*\"function\" + 0.005*\"cases\" + 0.005*\"value\" + 0.004*\"bit\" + 0.004*\"list\" + 0.004*\"looks like\" + 0.004*\"error\" + 0.004*\"does\" + 0.003*\"different\" + 0.003*\"change\" + 0.003*\"object\" + 0.003*\"quite\" + 0.003*\"just\" + 0.003*\"using\" + 0.003*\"non\" + 0.003*\"return\" + 0.003*\"add\" + 0.003*\"storage\" + 0.003*\"look\" + 0.003*\"method\" + 0.002*\"instead\" + 0.002*\"syntax\" + 0.002*\"specific\" + 0.002*\"null\" + 0.002*\"style\" + 0.002*\"functions\" + 0.002*\"examples\" + 0.002*\"set\" + 0.002*\"article\" + 0.002*\"single\" + 0.002*\"multiple\" + 0.002*\"database\" + 0.002*\"common\" + 0.002*\"pattern\" + 0.002*\"similar\" + 0.002*\"state\" + 0.002*\"int\" + 0.002*\"check\" + 0.002*\"information\" + 0.002*\"string\" + 0.002*\"class\" + 0.002*\"values\" + 0.002*\"copy\" + 0.002*\"sql\" + 0.002*\"tree\" + 0.002*\"thing\" + 0.002*\"input\" + 0.002*\"macros\" + 0.002*\"errors\" + 0.002*\"ve\" + 0.002*\"reference\" + 0.002*\"idea\" + 0.002*\"standard\" + 0.002*\"order\" + 0.002*\"editor\" + 0.002*\"special\" + 0.002*\"context\" + 0.002*\"things\" + 0.002*\"rules\" + 0.002*\"variable\" + 0.002*\"exactly\" + 0.002*\"problem\" + 0.002*\"methods\" + 0.002*\"hash\" + 0.002*\"names\" + 0.002*\"forgot\" + 0.002*\"results\" + 0.001*\"structure\" + 0.001*\"lists\" + 0.001*\"correct\" + 0.001*\"ve read\" + 0.001*\"called\" + 0.001*\"result\" + 0.001*\"usage\" + 0.001*\"objects\" + 0.001*\"model\" + 0.001*\"block\" + 0.001*\"macro\" + 0.001*\"binary\" + 0.001*\"guide\" + 0.001*\"oracle\" + 0.001*\"pretty\" + 0.001*\"actually\" + 0.001*\"bytecode\" + 0.001*\"intersection\" + 0.001*\"writes\" + 0.001*\"privacy\" + 0.001*\"defined\" + 0.001*\"feature\" + 0.001*\"handle\"\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[(0,\n",
       "  '0.008*\"companies\" + 0.008*\"uber\" + 0.008*\"cost\" + 0.008*\"company\" + 0.008*\"market\" + 0.008*\"business\" + 0.006*\"price\" + 0.006*\"people\" + 0.006*\"car\" + 0.005*\"long\" + 0.004*\"money\" + 0.004*\"tech\" + 0.004*\"tip\" + 0.004*\"high\" + 0.004*\"just\" + 0.004*\"self\" + 0.004*\"going\" + 0.004*\"buy\" + 0.003*\"term\" + 0.003*\"costs\" + 0.003*\"quality\" + 0.003*\"point\" + 0.003*\"small\" + 0.003*\"drivers\" + 0.003*\"probably\" + 0.003*\"big\" + 0.003*\"make\" + 0.003*\"model\" + 0.003*\"driving\" + 0.003*\"low\" + 0.003*\"industry\" + 0.003*\"cars\" + 0.003*\"tesla\" + 0.003*\"expensive\" + 0.003*\"time\" + 0.003*\"years\" + 0.003*\"like\" + 0.003*\"ve\" + 0.003*\"new\" + 0.003*\"million\" + 0.003*\"risk\" + 0.002*\"startup\" + 0.002*\"cheap\" + 0.002*\"prices\" + 0.002*\"stock\" + 0.002*\"supply\" + 0.002*\"large\" + 0.002*\"sell\" + 0.002*\"customers\" + 0.002*\"service\" + 0.002*\"value\" + 0.002*\"worth\" + 0.002*\"sales\" + 0.002*\"profit\" + 0.002*\"short\" + 0.002*\"startups\" + 0.002*\"businesses\" + 0.002*\"cheaper\" + 0.002*\"cash\" + 0.002*\"long term\" + 0.002*\"product\" + 0.002*\"lot\" + 0.002*\"meat\" + 0.002*\"good\" + 0.002*\"think\" + 0.002*\"twitter\" + 0.002*\"getting\" + 0.002*\"selling\" + 0.002*\"drive\" + 0.002*\"demand\" + 0.002*\"buying\" + 0.002*\"amazon\" + 0.002*\"making\" + 0.002*\"huge\" + 0.002*\"future\" + 0.002*\"investment\" + 0.002*\"pricing\" + 0.002*\"successful\" + 0.002*\"scale\" + 0.002*\"news\" + 0.002*\"tips\" + 0.002*\"medium\" + 0.002*\"products\" + 0.002*\"technology\" + 0.002*\"success\" + 0.001*\"competition\" + 0.001*\"millions\" + 0.001*\"status\" + 0.001*\"thing\" + 0.001*\"investors\" + 0.001*\"actually\" + 0.001*\"great\" + 0.001*\"growth\" + 0.001*\"hacker\" + 0.001*\"start\" + 0.001*\"dollars\" + 0.001*\"year\" + 0.001*\"right\" + 0.001*\"pretty\" + 0.001*\"day\"'),\n",
       " (1,\n",
       "  '0.025*\"people\" + 0.012*\"don\" + 0.009*\"think\" + 0.008*\"doesn\" + 0.008*\"say\" + 0.007*\"mean\" + 0.007*\"just\" + 0.007*\"point\" + 0.007*\"know\" + 0.007*\"does\" + 0.006*\"like\" + 0.005*\"question\" + 0.005*\"article\" + 0.005*\"person\" + 0.005*\"make\" + 0.005*\"understand\" + 0.005*\"actually\" + 0.004*\"way\" + 0.004*\"true\" + 0.004*\"fact\" + 0.004*\"wrong\" + 0.004*\"saying\" + 0.004*\"said\" + 0.004*\"believe\" + 0.004*\"evidence\" + 0.004*\"argument\" + 0.004*\"isn\" + 0.004*\"case\" + 0.003*\"human\" + 0.003*\"real\" + 0.003*\"didn\" + 0.003*\"different\" + 0.003*\"comment\" + 0.003*\"men\" + 0.003*\"means\" + 0.003*\"right\" + 0.003*\"reason\" + 0.003*\"talking\" + 0.003*\"sure\" + 0.003*\"did\" + 0.002*\"care\" + 0.002*\"things\" + 0.002*\"science\" + 0.002*\"clear\" + 0.002*\"number\" + 0.002*\"women\" + 0.002*\"answer\" + 0.002*\"really\" + 0.002*\"read\" + 0.002*\"matter\" + 0.002*\"example\" + 0.002*\"words\" + 0.002*\"given\" + 0.002*\"non\" + 0.002*\"value\" + 0.002*\"yes\" + 0.002*\"simply\" + 0.002*\"claim\" + 0.002*\"says\" + 0.002*\"credit\" + 0.002*\"reality\" + 0.002*\"good\" + 0.002*\"based\" + 0.002*\"word\" + 0.002*\"humans\" + 0.002*\"black\" + 0.002*\"aren\" + 0.002*\"agree\" + 0.002*\"exist\" + 0.002*\"change\" + 0.002*\"having\" + 0.002*\"definition\" + 0.002*\"response\" + 0.002*\"possible\" + 0.002*\"explain\" + 0.002*\"view\" + 0.002*\"mb\" + 0.002*\"statement\" + 0.002*\"argue\" + 0.002*\"correct\" + 0.002*\"don know\" + 0.002*\"study\" + 0.002*\"particular\" + 0.002*\"difference\" + 0.002*\"clearly\" + 0.002*\"discussion\" + 0.002*\"meaning\" + 0.002*\"better\" + 0.002*\"doesn mean\" + 0.002*\"general\" + 0.001*\"context\" + 0.001*\"reasonable\" + 0.001*\"life\" + 0.001*\"thought\" + 0.001*\"man\" + 0.001*\"truth\" + 0.001*\"obvious\" + 0.001*\"theory\" + 0.001*\"don think\" + 0.001*\"don understand\"'),\n",
       " (2,\n",
       "  '0.020*\"google\" + 0.012*\"open\" + 0.010*\"content\" + 0.009*\"free\" + 0.009*\"source\" + 0.009*\"don\" + 0.009*\"software\" + 0.008*\"product\" + 0.008*\"users\" + 0.008*\"make\" + 0.007*\"internet\" + 0.007*\"use\" + 0.007*\"facebook\" + 0.006*\"service\" + 0.006*\"want\" + 0.006*\"open source\" + 0.006*\"ads\" + 0.006*\"site\" + 0.006*\"just\" + 0.005*\"like\" + 0.005*\"security\" + 0.005*\"company\" + 0.005*\"network\" + 0.004*\"ad\" + 0.004*\"license\" + 0.004*\"sites\" + 0.004*\"email\" + 0.004*\"money\" + 0.004*\"need\" + 0.004*\"access\" + 0.004*\"business\" + 0.003*\"time\" + 0.003*\"nintendo\" + 0.003*\"app\" + 0.003*\"companies\" + 0.003*\"online\" + 0.003*\"devices\" + 0.003*\"phone\" + 0.003*\"gpl\" + 0.003*\"web\" + 0.003*\"services\" + 0.003*\"doesn\" + 0.003*\"advertising\" + 0.003*\"good\" + 0.003*\"project\" + 0.003*\"know\" + 0.003*\"device\" + 0.003*\"pay\" + 0.002*\"products\" + 0.002*\"user\" + 0.002*\"sell\" + 0.002*\"lock\" + 0.002*\"new\" + 0.002*\"let\" + 0.002*\"llc\" + 0.002*\"idea\" + 0.002*\"support\" + 0.002*\"marketing\" + 0.002*\"sure\" + 0.002*\"people\" + 0.002*\"networks\" + 0.002*\"account\" + 0.002*\"big\" + 0.002*\"angular\" + 0.002*\"ll\" + 0.002*\"website\" + 0.002*\"isn\" + 0.002*\"projects\" + 0.002*\"provide\" + 0.002*\"paid\" + 0.002*\"available\" + 0.002*\"microsoft\" + 0.002*\"did\" + 0.002*\"freebsd\" + 0.002*\"commercial\" + 0.002*\"bitcoin\" + 0.002*\"making\" + 0.002*\"won\" + 0.002*\"fake\" + 0.002*\"digital\" + 0.002*\"work\" + 0.002*\"tracking\" + 0.002*\"upgrade\" + 0.002*\"slack\" + 0.002*\"going\" + 0.002*\"community\" + 0.002*\"using\" + 0.002*\"search\" + 0.001*\"personal\" + 0.001*\"way\" + 0.001*\"founder\" + 0.001*\"websites\" + 0.001*\"offer\" + 0.001*\"non\" + 0.001*\"block\" + 0.001*\"messages\" + 0.001*\"smartphone\" + 0.001*\"customers\" + 0.001*\"drm\" + 0.001*\"create\"'),\n",
       " (3,\n",
       "  '0.013*\"use\" + 0.008*\"just\" + 0.007*\"user\" + 0.007*\"information\" + 0.006*\"does\" + 0.006*\"want\" + 0.006*\"traffic\" + 0.005*\"android\" + 0.005*\"phone\" + 0.005*\"using\" + 0.005*\"run\" + 0.005*\"pulseaudio\" + 0.004*\"access\" + 0.004*\"audio\" + 0.004*\"service\" + 0.004*\"need\" + 0.004*\"don\" + 0.004*\"ve\" + 0.004*\"share\" + 0.004*\"ip\" + 0.004*\"alsa\" + 0.004*\"data\" + 0.004*\"linux\" + 0.003*\"account\" + 0.003*\"desktop\" + 0.003*\"kernel\" + 0.003*\"different\" + 0.003*\"click\" + 0.003*\"doesn\" + 0.003*\"apng\" + 0.003*\"sure\" + 0.003*\"services\" + 0.003*\"like\" + 0.003*\"format\" + 0.003*\"key\" + 0.003*\"work\" + 0.003*\"ios\" + 0.003*\"address\" + 0.003*\"home\" + 0.003*\"case\" + 0.003*\"number\" + 0.002*\"having\" + 0.002*\"switch\" + 0.002*\"running\" + 0.002*\"netflix\" + 0.002*\"cloud\" + 0.002*\"party\" + 0.002*\"email\" + 0.002*\"bike\" + 0.002*\"tried\" + 0.002*\"trust\" + 0.002*\"aws\" + 0.002*\"password\" + 0.002*\"multiple\" + 0.002*\"signal\" + 0.002*\"app\" + 0.002*\"gif\" + 0.002*\"tv\" + 0.002*\"host\" + 0.002*\"connection\" + 0.002*\"way\" + 0.002*\"completely\" + 0.002*\"freedom\" + 0.002*\"able\" + 0.002*\"network\" + 0.002*\"cross\" + 0.002*\"info\" + 0.002*\"wouldn\" + 0.002*\"sent\" + 0.002*\"request\" + 0.002*\"scroll\" + 0.002*\"isp\" + 0.002*\"used\" + 0.002*\"mobile\" + 0.002*\"video\" + 0.002*\"voice\" + 0.002*\"time\" + 0.002*\"provider\" + 0.002*\"isps\" + 0.002*\"public\" + 0.002*\"single\" + 0.002*\"platform\" + 0.002*\"docker\" + 0.002*\"mail\" + 0.002*\"sign\" + 0.002*\"hosting\" + 0.002*\"make\" + 0.002*\"fb\" + 0.002*\"player\" + 0.002*\"set\" + 0.002*\"debian\" + 0.002*\"computer\" + 0.002*\"send\" + 0.002*\"hardware\" + 0.002*\"control\" + 0.002*\"music\" + 0.002*\"know\" + 0.002*\"store\" + 0.001*\"accounts\" + 0.001*\"lot\"'),\n",
       " (4,\n",
       "  '0.011*\"life\" + 0.008*\"post\" + 0.007*\"like\" + 0.006*\"hn\" + 0.006*\"people\" + 0.006*\"ve\" + 0.006*\"social\" + 0.005*\"title\" + 0.004*\"friends\" + 0.004*\"didn\" + 0.004*\"got\" + 0.004*\"thing\" + 0.004*\"search\" + 0.004*\"comment\" + 0.004*\"point\" + 0.004*\"just\" + 0.004*\"pro\" + 0.003*\"read\" + 0.003*\"commit\" + 0.003*\"happy\" + 0.003*\"link\" + 0.003*\"right\" + 0.003*\"think\" + 0.003*\"did\" + 0.003*\"movie\" + 0.003*\"story\" + 0.003*\"way\" + 0.003*\"thanks\" + 0.003*\"privacy\" + 0.003*\"oh\" + 0.003*\"battery\" + 0.003*\"kids\" + 0.003*\"laptop\" + 0.003*\"family\" + 0.003*\"interesting\" + 0.003*\"parent\" + 0.003*\"maybe\" + 0.002*\"ratings\" + 0.002*\"thought\" + 0.002*\"comments\" + 0.002*\"new\" + 0.002*\"away\" + 0.002*\"film\" + 0.002*\"media\" + 0.002*\"yes\" + 0.002*\"thread\" + 0.002*\"probably\" + 0.002*\"stories\" + 0.002*\"article\" + 0.002*\"ask\" + 0.002*\"bsd\" + 0.002*\"seen\" + 0.002*\"treatment\" + 0.002*\"audience\" + 0.002*\"kind\" + 0.002*\"guess\" + 0.002*\"don\" + 0.002*\"stop\" + 0.002*\"community\" + 0.002*\"child\" + 0.002*\"hope\" + 0.002*\"ll\" + 0.002*\"blog post\" + 0.002*\"star\" + 0.002*\"said\" + 0.002*\"time\" + 0.002*\"person\" + 0.002*\"ah\" + 0.002*\"movies\" + 0.002*\"aware\" + 0.002*\"wasn\" + 0.002*\"know\" + 0.002*\"good\" + 0.002*\"old\" + 0.002*\"original\" + 0.002*\"day\" + 0.002*\"defend\" + 0.002*\"networking\" + 0.001*\"sure\" + 0.001*\"page\" + 0.001*\"cells\" + 0.001*\"mention\" + 0.001*\"calling\" + 0.001*\"looking\" + 0.001*\"arrow\" + 0.001*\"care\" + 0.001*\"agents\" + 0.001*\"children\" + 0.001*\"sorry\" + 0.001*\"interaction\" + 0.001*\"agent\" + 0.001*\"edit\" + 0.001*\"seeing\" + 0.001*\"self\" + 0.001*\"proof\" + 0.001*\"editing\" + 0.001*\"reading\" + 0.001*\"links\" + 0.001*\"phrase\" + 0.001*\"taken\"'),\n",
       " (5,\n",
       "  '0.023*\"https\" + 0.016*\"org\" + 0.013*\"github\" + 0.011*\"windows\" + 0.011*\"wikipedia\" + 0.011*\"web\" + 0.009*\"en\" + 0.009*\"wiki\" + 0.009*\"com\" + 0.009*\"use\" + 0.008*\"github com\" + 0.008*\"org wiki\" + 0.008*\"wikipedia org\" + 0.008*\"https github\" + 0.008*\"https en\" + 0.008*\"en wikipedia\" + 0.007*\"browser\" + 0.007*\"linux\" + 0.007*\"using\" + 0.007*\"page\" + 0.006*\"server\" + 0.006*\"support\" + 0.006*\"file\" + 0.006*\"apps\" + 0.005*\"react\" + 0.005*\"app\" + 0.005*\"html\" + 0.005*\"gitlab\" + 0.005*\"version\" + 0.004*\"os\" + 0.004*\"text\" + 0.004*\"install\" + 0.004*\"just\" + 0.004*\"like\" + 0.004*\"issues\" + 0.004*\"microsoft\" + 0.004*\"files\" + 0.004*\"chrome\" + 0.004*\"firefox\" + 0.003*\"git\" + 0.003*\"io\" + 0.003*\"used\" + 0.003*\"http\" + 0.003*\"user\" + 0.003*\"ms\" + 0.003*\"running\" + 0.003*\"css\" + 0.003*\"native\" + 0.003*\"ui\" + 0.003*\"open\" + 0.003*\"release\" + 0.003*\"client\" + 0.003*\"download\" + 0.003*\"docs\" + 0.002*\"works\" + 0.002*\"mac\" + 0.002*\"default\" + 0.002*\"update\" + 0.002*\"browsers\" + 0.002*\"source\" + 0.002*\"updates\" + 0.002*\"master\" + 0.002*\"json\" + 0.002*\"run\" + 0.002*\"window\" + 0.002*\"does\" + 0.002*\"work\" + 0.002*\"add\" + 0.002*\"aframe\" + 0.002*\"bar\" + 0.002*\"jpeg\" + 0.002*\"api\" + 0.002*\"supported\" + 0.002*\"new\" + 0.002*\"nice\" + 0.002*\"users\" + 0.002*\"module\" + 0.002*\"command\" + 0.002*\"edit\" + 0.002*\"display\" + 0.002*\"site\" + 0.002*\"screen\" + 0.002*\"list\" + 0.002*\"stack\" + 0.002*\"unix\" + 0.002*\"pages\" + 0.002*\"keys\" + 0.002*\"link\" + 0.002*\"emacs\" + 0.002*\"web components\" + 0.002*\"blob\" + 0.002*\"keyboard\" + 0.001*\"based\" + 0.001*\"archive\" + 0.001*\"terminal\" + 0.001*\"released\" + 0.001*\"mobile\" + 0.001*\"example\" + 0.001*\"delaware\" + 0.001*\"roundup\"'),\n",
       " (6,\n",
       "  '0.021*\"don\" + 0.018*\"just\" + 0.018*\"think\" + 0.017*\"like\" + 0.015*\"really\" + 0.012*\"people\" + 0.010*\"work\" + 0.009*\"need\" + 0.009*\"know\" + 0.008*\"way\" + 0.008*\"time\" + 0.008*\"want\" + 0.007*\"make\" + 0.007*\"ve\" + 0.007*\"things\" + 0.006*\"lot\" + 0.005*\"good\" + 0.005*\"feel\" + 0.005*\"makes\" + 0.005*\"better\" + 0.005*\"doing\" + 0.005*\"hard\" + 0.004*\"thing\" + 0.004*\"going\" + 0.004*\"sense\" + 0.004*\"learn\" + 0.004*\"experience\" + 0.004*\"ll\" + 0.003*\"programming\" + 0.003*\"don think\" + 0.003*\"actually\" + 0.003*\"stuff\" + 0.003*\"probably\" + 0.003*\"years\" + 0.003*\"end\" + 0.003*\"understand\" + 0.003*\"day\" + 0.003*\"don know\" + 0.003*\"new\" + 0.003*\"learning\" + 0.003*\"right\" + 0.003*\"working\" + 0.002*\"having\" + 0.002*\"point\" + 0.002*\"math\" + 0.002*\"bit\" + 0.002*\"course\" + 0.002*\"little\" + 0.002*\"language\" + 0.002*\"job\" + 0.002*\"look\" + 0.002*\"kind\" + 0.002*\"doesn\" + 0.002*\"trying\" + 0.002*\"best\" + 0.002*\"maybe\" + 0.002*\"come\" + 0.002*\"start\" + 0.002*\"level\" + 0.002*\"help\" + 0.002*\"getting\" + 0.002*\"say\" + 0.002*\"great\" + 0.002*\"sure\" + 0.002*\"different\" + 0.002*\"scala\" + 0.002*\"able\" + 0.002*\"pretty\" + 0.002*\"try\" + 0.002*\"idea\" + 0.002*\"isn\" + 0.002*\"looking\" + 0.002*\"feel like\" + 0.002*\"skills\" + 0.002*\"book\" + 0.002*\"agree\" + 0.002*\"easy\" + 0.002*\"started\" + 0.001*\"worked\" + 0.001*\"makes sense\" + 0.001*\"knowledge\" + 0.001*\"important\" + 0.001*\"components\" + 0.001*\"love\" + 0.001*\"easier\" + 0.001*\"thinking\" + 0.001*\"simple\" + 0.001*\"mind\" + 0.001*\"got\" + 0.001*\"quite\" + 0.001*\"especially\" + 0.001*\"advice\" + 0.001*\"school\" + 0.001*\"world\" + 0.001*\"common\" + 0.001*\"problem\" + 0.001*\"making\" + 0.001*\"ideas\" + 0.001*\"don want\" + 0.001*\"instead\"'),\n",
       " (7,\n",
       "  '0.008*\"women\" + 0.008*\"did\" + 0.007*\"war\" + 0.006*\"history\" + 0.005*\"patent\" + 0.004*\"later\" + 0.004*\"author\" + 0.004*\"read\" + 0.004*\"years\" + 0.004*\"wasn\" + 0.004*\"art\" + 0.004*\"corp\" + 0.004*\"man\" + 0.004*\"say\" + 0.003*\"remember\" + 0.003*\"russia\" + 0.003*\"female\" + 0.003*\"japanese\" + 0.003*\"patents\" + 0.003*\"culture\" + 0.003*\"lyft\" + 0.003*\"didn\" + 0.003*\"world\" + 0.003*\"reading\" + 0.003*\"just\" + 0.003*\"said\" + 0.003*\"comments\" + 0.003*\"china\" + 0.003*\"guy\" + 0.002*\"came\" + 0.002*\"english\" + 0.002*\"god\" + 0.002*\"reddit\" + 0.002*\"male\" + 0.002*\"th\" + 0.002*\"notes\" + 0.002*\"yep\" + 0.002*\"russian\" + 0.002*\"today\" + 0.002*\"comment\" + 0.002*\"dr\" + 0.002*\"recall\" + 0.002*\"totally\" + 0.002*\"likely\" + 0.002*\"usd\" + 0.002*\"restaurant\" + 0.002*\"chinese\" + 0.002*\"remain\" + 0.002*\"wonder\" + 0.002*\"time\" + 0.002*\"gun\" + 0.002*\"original\" + 0.002*\"like\" + 0.002*\"woman\" + 0.002*\"late\" + 0.002*\"fuck\" + 0.002*\"story\" + 0.002*\"early\" + 0.002*\"coffee\" + 0.002*\"mit\" + 0.002*\"thought\" + 0.002*\"right\" + 0.002*\"got\" + 0.002*\"years later\" + 0.002*\"bias\" + 0.002*\"sum\" + 0.002*\"databases\" + 0.002*\"labour\" + 0.002*\"restrictions\" + 0.002*\"gold\" + 0.001*\"codecs\" + 0.001*\"macos\" + 0.001*\"directed\" + 0.001*\"neat\" + 0.001*\"entertainment\" + 0.001*\"section\" + 0.001*\"analytics\" + 0.001*\"origin\" + 0.001*\"going\" + 0.001*\"tl\" + 0.001*\"paper\" + 0.001*\"long\" + 0.001*\"imbalance\" + 0.001*\"internet\" + 0.001*\"onions\" + 0.001*\"agreed\" + 0.001*\"tl dr\" + 0.001*\"sticker\" + 0.001*\"honest\" + 0.001*\"headline\" + 0.001*\"mind\" + 0.001*\"pre\" + 0.001*\"century\" + 0.001*\"ass\" + 0.001*\"artificial\" + 0.001*\"referring\" + 0.001*\"sex\" + 0.001*\"far\" + 0.001*\"shock\" + 0.001*\"true\"'),\n",
       " (8,\n",
       "  '0.009*\"government\" + 0.009*\"law\" + 0.007*\"right\" + 0.006*\"state\" + 0.006*\"power\" + 0.005*\"people\" + 0.005*\"trump\" + 0.005*\"like\" + 0.005*\"bad\" + 0.004*\"public\" + 0.004*\"legal\" + 0.004*\"states\" + 0.004*\"political\" + 0.004*\"society\" + 0.003*\"laws\" + 0.003*\"age\" + 0.003*\"court\" + 0.003*\"policy\" + 0.003*\"think\" + 0.003*\"world\" + 0.003*\"way\" + 0.003*\"risk\" + 0.003*\"vote\" + 0.002*\"party\" + 0.002*\"dating\" + 0.002*\"rights\" + 0.002*\"free\" + 0.002*\"anti\" + 0.002*\"rules\" + 0.002*\"want\" + 0.002*\"thing\" + 0.002*\"social\" + 0.002*\"just\" + 0.002*\"best\" + 0.002*\"interests\" + 0.002*\"act\" + 0.002*\"good\" + 0.002*\"current\" + 0.002*\"change\" + 0.002*\"idea\" + 0.002*\"isn\" + 0.002*\"regulation\" + 0.002*\"case\" + 0.002*\"place\" + 0.002*\"united\" + 0.002*\"majority\" + 0.002*\"american\" + 0.002*\"economic\" + 0.002*\"win\" + 0.002*\"views\" + 0.002*\"moral\" + 0.002*\"kind\" + 0.002*\"likely\" + 0.002*\"possible\" + 0.002*\"voting\" + 0.002*\"regulations\" + 0.002*\"does\" + 0.002*\"private\" + 0.002*\"result\" + 0.002*\"wing\" + 0.002*\"human\" + 0.002*\"united states\" + 0.002*\"control\" + 0.001*\"let\" + 0.001*\"decision\" + 0.001*\"agree\" + 0.001*\"effect\" + 0.001*\"federal\" + 0.001*\"left\" + 0.001*\"san\" + 0.001*\"order\" + 0.001*\"means\" + 0.001*\"citizens\" + 0.001*\"issue\" + 0.001*\"net\" + 0.001*\"position\" + 0.001*\"making\" + 0.001*\"certain\" + 0.001*\"era\" + 0.001*\"fight\" + 0.001*\"rule\" + 0.001*\"individual\" + 0.001*\"example\" + 0.001*\"important\" + 0.001*\"policies\" + 0.001*\"property\" + 0.001*\"fly\" + 0.001*\"orders\" + 0.001*\"chance\" + 0.001*\"defense\" + 0.001*\"better\" + 0.001*\"market\" + 0.001*\"action\" + 0.001*\"support\" + 0.001*\"wouldn\" + 0.001*\"research\" + 0.001*\"democracy\" + 0.001*\"financial\" + 0.001*\"given\" + 0.001*\"culture\"'),\n",
       " (9,\n",
       "  '0.014*\"pay\" + 0.013*\"money\" + 0.010*\"work\" + 0.009*\"job\" + 0.009*\"people\" + 0.008*\"year\" + 0.007*\"jobs\" + 0.007*\"tax\" + 0.005*\"item\" + 0.005*\"rate\" + 0.005*\"paying\" + 0.005*\"home\" + 0.005*\"income\" + 0.005*\"ycombinator\" + 0.005*\"paid\" + 0.005*\"company\" + 0.005*\"country\" + 0.005*\"ycombinator com\" + 0.005*\"id\" + 0.005*\"https news\" + 0.005*\"living\" + 0.005*\"news ycombinator\" + 0.005*\"news\" + 0.004*\"time\" + 0.004*\"taxes\" + 0.004*\"hours\" + 0.004*\"item id\" + 0.004*\"com item\" + 0.004*\"live\" + 0.004*\"tipping\" + 0.004*\"debt\" + 0.004*\"don\" + 0.004*\"countries\" + 0.003*\"working\" + 0.003*\"driver\" + 0.003*\"health\" + 0.003*\"higher\" + 0.003*\"high\" + 0.003*\"insurance\" + 0.003*\"school\" + 0.003*\"government\" + 0.003*\"cost\" + 0.003*\"house\" + 0.003*\"poor\" + 0.003*\"education\" + 0.003*\"lower\" + 0.003*\"city\" + 0.003*\"month\" + 0.003*\"workers\" + 0.003*\"employees\" + 0.003*\"uk\" + 0.002*\"healthcare\" + 0.002*\"salary\" + 0.002*\"companies\" + 0.002*\"rates\" + 0.002*\"employee\" + 0.002*\"wage\" + 0.002*\"labor\" + 0.002*\"years\" + 0.002*\"rural\" + 0.002*\"college\" + 0.002*\"day\" + 0.002*\"class\" + 0.002*\"rating\" + 0.002*\"office\" + 0.002*\"employer\" + 0.002*\"average\" + 0.002*\"costs\" + 0.002*\"care\" + 0.002*\"getting\" + 0.002*\"spend\" + 0.002*\"america\" + 0.002*\"minimum\" + 0.002*\"hire\" + 0.002*\"family\" + 0.002*\"rent\" + 0.002*\"week\" + 0.002*\"lot\" + 0.002*\"kids\" + 0.002*\"local\" + 0.002*\"rich\" + 0.002*\"leave\" + 0.002*\"just\" + 0.002*\"low\" + 0.002*\"afford\" + 0.002*\"life\" + 0.002*\"area\" + 0.002*\"housing\" + 0.002*\"students\" + 0.002*\"capital\" + 0.002*\"places\" + 0.002*\"schools\" + 0.002*\"american\" + 0.002*\"million\" + 0.002*\"public\" + 0.002*\"wages\" + 0.002*\"middle\" + 0.002*\"remote\" + 0.002*\"free\" + 0.002*\"spending\"'),\n",
       " (10,\n",
       "  '0.021*\"code\" + 0.013*\"use\" + 0.007*\"java\" + 0.007*\"language\" + 0.006*\"using\" + 0.006*\"write\" + 0.004*\"memory\" + 0.004*\"systems\" + 0.004*\"time\" + 0.004*\"used\" + 0.004*\"like\" + 0.004*\"machine\" + 0.004*\"run\" + 0.003*\"build\" + 0.003*\"js\" + 0.003*\"need\" + 0.003*\"performance\" + 0.003*\"simple\" + 0.003*\"project\" + 0.003*\"library\" + 0.003*\"rust\" + 0.003*\"python\" + 0.003*\"design\" + 0.003*\"just\" + 0.003*\"easy\" + 0.003*\"tools\" + 0.003*\"features\" + 0.003*\"level\" + 0.003*\"does\" + 0.003*\"javascript\" + 0.003*\"writing\" + 0.003*\"written\" + 0.003*\"learning\" + 0.003*\"way\" + 0.003*\"api\" + 0.002*\"compiler\" + 0.002*\"program\" + 0.002*\"languages\" + 0.002*\"developers\" + 0.002*\"software\" + 0.002*\"really\" + 0.002*\"implementation\" + 0.002*\"new\" + 0.002*\"standard\" + 0.002*\"running\" + 0.002*\"development\" + 0.002*\"based\" + 0.002*\"lot\" + 0.002*\"example\" + 0.002*\"different\" + 0.002*\"hardware\" + 0.002*\"built\" + 0.002*\"support\" + 0.002*\"image\" + 0.002*\"application\" + 0.002*\"want\" + 0.002*\"possible\" + 0.002*\"great\" + 0.002*\"small\" + 0.002*\"libraries\" + 0.002*\"fast\" + 0.002*\"source\" + 0.002*\"faster\" + 0.002*\"better\" + 0.002*\"framework\" + 0.002*\"pretty\" + 0.002*\"easier\" + 0.002*\"vs\" + 0.002*\"line\" + 0.002*\"nice\" + 0.002*\"lines\" + 0.002*\"actually\" + 0.002*\"feature\" + 0.002*\"applications\" + 0.002*\"make\" + 0.002*\"set\" + 0.002*\"size\" + 0.002*\"stuff\" + 0.002*\"uses\" + 0.002*\"useful\" + 0.002*\"interesting\" + 0.002*\"dom\" + 0.002*\"core\" + 0.002*\"cpu\" + 0.002*\"test\" + 0.002*\"architecture\" + 0.002*\"single\" + 0.002*\"designed\" + 0.002*\"complex\" + 0.002*\"able\" + 0.002*\"end\" + 0.002*\"speed\" + 0.002*\"haskell\" + 0.002*\"implement\" + 0.002*\"compatible\" + 0.002*\"static\" + 0.002*\"source code\" + 0.002*\"instead\" + 0.002*\"model\" + 0.002*\"load\"'),\n",
       " (11,\n",
       "  '0.006*\"water\" + 0.005*\"light\" + 0.005*\"energy\" + 0.005*\"power\" + 0.004*\"police\" + 0.004*\"food\" + 0.004*\"city\" + 0.003*\"world\" + 0.003*\"fiber\" + 0.003*\"space\" + 0.003*\"air\" + 0.003*\"monsanto\" + 0.003*\"papers\" + 0.003*\"mass\" + 0.003*\"cold\" + 0.003*\"red\" + 0.003*\"new\" + 0.003*\"large\" + 0.002*\"like\" + 0.002*\"hole\" + 0.002*\"berkeley\" + 0.002*\"area\" + 0.002*\"pdf\" + 0.002*\"far\" + 0.002*\"earth\" + 0.002*\"global\" + 0.002*\"high\" + 0.002*\"drug\" + 0.002*\"times\" + 0.002*\"humanity\" + 0.002*\"military\" + 0.002*\"live\" + 0.002*\"article\" + 0.002*\"flight\" + 0.002*\"pressure\" + 0.002*\"nuclear\" + 0.002*\"body\" + 0.002*\"glyphosate\" + 0.002*\"population\" + 0.002*\"crime\" + 0.002*\"safe\" + 0.002*\"black\" + 0.002*\"small\" + 0.002*\"paper\" + 0.002*\"weight\" + 0.002*\"eat\" + 0.002*\"just\" + 0.002*\"land\" + 0.002*\"weapons\" + 0.002*\"budget\" + 0.002*\"cities\" + 0.002*\"australia\" + 0.002*\"effects\" + 0.002*\"cause\" + 0.002*\"black hole\" + 0.002*\"areas\" + 0.002*\"dangerous\" + 0.002*\"climate\" + 0.002*\"heat\" + 0.002*\"research\" + 0.002*\"speed\" + 0.002*\"scale\" + 0.002*\"quite\" + 0.002*\"place\" + 0.002*\"ideal\" + 0.002*\"near\" + 0.002*\"south\" + 0.002*\"aircraft\" + 0.002*\"comma\" + 0.002*\"case\" + 0.001*\"fear\" + 0.001*\"away\" + 0.001*\"human\" + 0.001*\"low\" + 0.001*\"lot\" + 0.001*\"german\" + 0.001*\"normal\" + 0.001*\"ground\" + 0.001*\"temperature\" + 0.001*\"north\" + 0.001*\"possible\" + 0.001*\"york\" + 0.001*\"inside\" + 0.001*\"feet\" + 0.001*\"new york\" + 0.001*\"offline\" + 0.001*\"kill\" + 0.001*\"plane\" + 0.001*\"close\" + 0.001*\"likely\" + 0.001*\"obama\" + 0.001*\"pretty\" + 0.001*\"traffic light\" + 0.001*\"probably\" + 0.001*\"french\" + 0.001*\"big\" + 0.001*\"safety\" + 0.001*\"produce\" + 0.001*\"glass\" + 0.001*\"bit\"'),\n",
       " (12,\n",
       "  '0.052*\"com\" + 0.038*\"https\" + 0.033*\"www\" + 0.029*\"http\" + 0.018*\"https www\" + 0.014*\"http www\" + 0.012*\"ago\" + 0.011*\"video\" + 0.011*\"years\" + 0.009*\"youtube\" + 0.009*\"news\" + 0.008*\"watch\" + 0.007*\"google\" + 0.007*\"like\" + 0.007*\"blog\" + 0.006*\"years ago\" + 0.006*\"org\" + 0.005*\"videos\" + 0.004*\"apple\" + 0.004*\"kotlin\" + 0.004*\"article\" + 0.004*\"amazon\" + 0.004*\"great\" + 0.004*\"html\" + 0.003*\"book\" + 0.003*\"youtube com\" + 0.003*\"sounds\" + 0.003*\"articles\" + 0.003*\"www youtube\" + 0.003*\"com watch\" + 0.003*\"year\" + 0.003*\"new\" + 0.003*\"day\" + 0.003*\"bay\" + 0.003*\"interesting\" + 0.002*\"yc\" + 0.002*\"google com\" + 0.002*\"net\" + 0.002*\"months\" + 0.002*\"link\" + 0.002*\"sounds like\" + 0.002*\"cloud\" + 0.002*\"aws\" + 0.002*\"looking\" + 0.002*\"gaming\" + 0.002*\"couple\" + 0.002*\"bay area\" + 0.002*\"vim\" + 0.002*\"recommend\" + 0.002*\"camera\" + 0.002*\"site\" + 0.002*\"vr\" + 0.002*\"days\" + 0.002*\"saw\" + 0.002*\"gov\" + 0.002*\"png\" + 0.002*\"page\" + 0.002*\"recently\" + 0.002*\"ipad\" + 0.002*\"time\" + 0.002*\"dropbox\" + 0.002*\"books\" + 0.002*\"edu\" + 0.002*\"ve\" + 0.002*\"archive\" + 0.002*\"launch\" + 0.002*\"bought\" + 0.002*\"week\" + 0.002*\"providers\" + 0.002*\"amazing\" + 0.002*\"good\" + 0.002*\"reading\" + 0.002*\"used\" + 0.002*\"similar\" + 0.002*\"designer\" + 0.002*\"lyrics\" + 0.002*\"jack\" + 0.001*\"drive\" + 0.001*\"search\" + 0.001*\"channel\" + 0.001*\"games\" + 0.001*\"looks\" + 0.001*\"post\" + 0.001*\"linear\" + 0.001*\"remember\" + 0.001*\"topic\" + 0.001*\"recent\" + 0.001*\"uk\" + 0.001*\"look\" + 0.001*\"gb\" + 0.001*\"check\" + 0.001*\"edit\" + 0.001*\"interested\" + 0.001*\"did\" + 0.001*\"pretty\" + 0.001*\"read\" + 0.001*\"story\" + 0.001*\"series\" + 0.001*\"got\" + 0.001*\"best\"'),\n",
       " (13,\n",
       "  '0.015*\"time\" + 0.014*\"problem\" + 0.012*\"good\" + 0.009*\"game\" + 0.009*\"ve\" + 0.008*\"things\" + 0.007*\"just\" + 0.007*\"problems\" + 0.007*\"lot\" + 0.007*\"years\" + 0.006*\"better\" + 0.006*\"games\" + 0.005*\"real\" + 0.005*\"like\" + 0.005*\"old\" + 0.005*\"work\" + 0.005*\"long\" + 0.005*\"thanks\" + 0.004*\"ll\" + 0.004*\"interview\" + 0.004*\"ai\" + 0.004*\"new\" + 0.004*\"sure\" + 0.004*\"questions\" + 0.004*\"maybe\" + 0.003*\"great\" + 0.003*\"solve\" + 0.003*\"software\" + 0.003*\"issues\" + 0.003*\"going\" + 0.003*\"thing\" + 0.003*\"world\" + 0.003*\"know\" + 0.003*\"play\" + 0.003*\"pretty\" + 0.003*\"way\" + 0.003*\"test\" + 0.003*\"year\" + 0.003*\"team\" + 0.003*\"really\" + 0.003*\"whiteboard\" + 0.003*\"actually\" + 0.003*\"tell\" + 0.003*\"bad\" + 0.003*\"issue\" + 0.003*\"solution\" + 0.003*\"did\" + 0.003*\"fix\" + 0.003*\"testing\" + 0.003*\"isn\" + 0.002*\"heard\" + 0.002*\"haven\" + 0.002*\"tests\" + 0.002*\"try\" + 0.002*\"definitely\" + 0.002*\"talk\" + 0.002*\"quality\" + 0.002*\"kind\" + 0.002*\"best\" + 0.002*\"process\" + 0.002*\"big\" + 0.002*\"having\" + 0.002*\"long time\" + 0.002*\"far\" + 0.002*\"got\" + 0.002*\"didn\" + 0.002*\"fun\" + 0.002*\"don\" + 0.002*\"trying\" + 0.002*\"engineering\" + 0.002*\"working\" + 0.002*\"different\" + 0.002*\"think\" + 0.002*\"candidate\" + 0.002*\"right\" + 0.002*\"idea\" + 0.002*\"seen\" + 0.002*\"luck\" + 0.002*\"playing\" + 0.002*\"thank\" + 0.002*\"hard\" + 0.002*\"probably\" + 0.002*\"algorithm\" + 0.002*\"looking\" + 0.002*\"answer\" + 0.002*\"interviews\" + 0.002*\"start\" + 0.002*\"help\" + 0.002*\"people\" + 0.002*\"technical\" + 0.002*\"cool\" + 0.002*\"developers\" + 0.002*\"works\" + 0.002*\"past\" + 0.002*\"ask\" + 0.001*\"mario\" + 0.001*\"doing\" + 0.001*\"solved\" + 0.001*\"change\" + 0.001*\"want\"'),\n",
       " (14,\n",
       "  '0.023*\"data\" + 0.017*\"like\" + 0.010*\"use\" + 0.008*\"type\" + 0.007*\"case\" + 0.007*\"used\" + 0.007*\"read\" + 0.006*\"looks\" + 0.006*\"types\" + 0.005*\"example\" + 0.005*\"function\" + 0.005*\"cases\" + 0.005*\"value\" + 0.004*\"bit\" + 0.004*\"list\" + 0.004*\"looks like\" + 0.004*\"error\" + 0.004*\"does\" + 0.003*\"different\" + 0.003*\"change\" + 0.003*\"object\" + 0.003*\"quite\" + 0.003*\"just\" + 0.003*\"using\" + 0.003*\"non\" + 0.003*\"return\" + 0.003*\"add\" + 0.003*\"storage\" + 0.003*\"look\" + 0.003*\"method\" + 0.002*\"instead\" + 0.002*\"syntax\" + 0.002*\"specific\" + 0.002*\"null\" + 0.002*\"style\" + 0.002*\"functions\" + 0.002*\"examples\" + 0.002*\"set\" + 0.002*\"article\" + 0.002*\"single\" + 0.002*\"multiple\" + 0.002*\"database\" + 0.002*\"common\" + 0.002*\"pattern\" + 0.002*\"similar\" + 0.002*\"state\" + 0.002*\"int\" + 0.002*\"check\" + 0.002*\"information\" + 0.002*\"string\" + 0.002*\"class\" + 0.002*\"values\" + 0.002*\"copy\" + 0.002*\"sql\" + 0.002*\"tree\" + 0.002*\"thing\" + 0.002*\"input\" + 0.002*\"macros\" + 0.002*\"errors\" + 0.002*\"ve\" + 0.002*\"reference\" + 0.002*\"idea\" + 0.002*\"standard\" + 0.002*\"order\" + 0.002*\"editor\" + 0.002*\"special\" + 0.002*\"context\" + 0.002*\"things\" + 0.002*\"rules\" + 0.002*\"variable\" + 0.002*\"exactly\" + 0.002*\"problem\" + 0.002*\"methods\" + 0.002*\"hash\" + 0.002*\"names\" + 0.002*\"forgot\" + 0.002*\"results\" + 0.001*\"structure\" + 0.001*\"lists\" + 0.001*\"correct\" + 0.001*\"ve read\" + 0.001*\"called\" + 0.001*\"result\" + 0.001*\"usage\" + 0.001*\"objects\" + 0.001*\"model\" + 0.001*\"block\" + 0.001*\"macro\" + 0.001*\"binary\" + 0.001*\"guide\" + 0.001*\"oracle\" + 0.001*\"pretty\" + 0.001*\"actually\" + 0.001*\"bytecode\" + 0.001*\"intersection\" + 0.001*\"writes\" + 0.001*\"privacy\" + 0.001*\"defined\" + 0.001*\"feature\" + 0.001*\"handle\"')]"
      ]
     },
     "execution_count": 117,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lda.print_topics(num_words = 100)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "New LDA-learned topics after regex:\n",
    "- Workplace banter (like, life, look, old, human, time, world, women, years)\n",
    "- Talks on open knowledge (read, article, source, open, open source, book, community)\n",
    "- Data talk (gooogle, data, ai, deep, intelligence, research, networks, connection, model)\n",
    "- Software performance (memory, speed, size, math, module, infinite, papers, hash)\n",
    "- Programming languages (code, language, java, programming, javascript, js, python, api, rust)\n",
    "- Macro business/govt (wikipedia, government, countries, income, taxes, power, state, country, capital)\n",
    "- Changes in history (years, ago, city, years ago, new, later, america, times, land, happened)\n",
    "- Big software platforms (linux, windows, app, software, web, android, microsoft, aws)\n",
    "- 'Uber' effect (money, cost, uber, market, car, problem, service, self, driving, drivers, tesla)\n",
    "- Generic remarks (point, better, things, great, experiment, best, question, answers)\n",
    "- Code management (data, user, image, file, server, computer, browser, linear, kernel, source\n",
    "- Government (law, government, legal, private, public, policy, issue, patent, trust, trump)\n",
    "- Job/family concerns (people, work, want, job, money, culture, problem, poor, family)\n",
    "- Popular website links/attachments (github, google, pdf, post, page, dropbox, reddit, gitlab)\n",
    "- Job search (experience, interview, learning, school, great, company, years, skills, whiteboard, startup, technical, questions)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This topics above are very insightful! Many have common buzzwords I've heard about the topics on Hacker News or from friends - but it's cool to all see them in one place!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The LDA and clustering used simple bag of words approaches - so we don't know the context of words in sentences. So when we heard words like \"good\", \"bad\" and \"cars\", we don't know exactly what people generally associate cars with. Let's try to make conceptual comparisions between blocks of text."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Word2vec**\n",
    "can create context windows like five-grams\n",
    "we need to assign each word or bigram an index from 0 - V  (v being # distinct words in corpus)\n",
    "Inputs: middle word\n",
    "Output: other words of context window\n",
    "\n",
    "Skip grams:\n",
    "We're trying to predict the words around a word.\n",
    "\n",
    "Continuous bag of words:\n",
    "given several words, can we predict the one-word associated with it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "stop = set(['heh', 'wow', 'hah', 'hey', 'yeah','wtf', 'meh', 'oops', 'nah', 'yea','doesnt','dont',\n",
    "   'huh', 'mar', 'umm', 'like', 'think', 'duh', 'sigh', 'wheres', 'hmm','interesting','article',\n",
    "   'say', 'hello', 'yup','im', 'ltsarcasmgt', 'hehe', 'blah', 'nope', 'ouch', 'uh'] + stopwords.words('english'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "186"
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(stop)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/dan/anaconda/lib/python3.6/site-packages/ipykernel/__main__.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  if __name__ == '__main__':\n"
     ]
    }
   ],
   "source": [
    "df_type_comments['split_comment_words'] = df_type_comments['text'].str.lower().str.split()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "texts = []\n",
    "\n",
    "for comment in df_type_comments.split_comment_words:\n",
    "    new_comment = []\n",
    "    for word in comment:\n",
    "        if word in stop:\n",
    "            pass  # we don't want to include this stop word\n",
    "        else:\n",
    "            new_comment.append(word)\n",
    "    texts.append(new_comment)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2017-06-05 21:01:38,173 : INFO : collecting all words and their counts\n",
      "2017-06-05 21:01:38,175 : INFO : PROGRESS: at sentence #0, processed 0 words, keeping 0 word types\n",
      "2017-06-05 21:01:38,262 : INFO : PROGRESS: at sentence #10000, processed 331136 words, keeping 53803 word types\n",
      "2017-06-05 21:01:38,348 : INFO : PROGRESS: at sentence #20000, processed 673855 words, keeping 85612 word types\n",
      "2017-06-05 21:01:38,435 : INFO : PROGRESS: at sentence #30000, processed 985195 words, keeping 110268 word types\n",
      "2017-06-05 21:01:38,526 : INFO : PROGRESS: at sentence #40000, processed 1323020 words, keeping 133862 word types\n",
      "2017-06-05 21:01:38,622 : INFO : PROGRESS: at sentence #50000, processed 1667684 words, keeping 156219 word types\n",
      "2017-06-05 21:01:38,729 : INFO : PROGRESS: at sentence #60000, processed 2006941 words, keeping 175991 word types\n",
      "2017-06-05 21:01:38,823 : INFO : PROGRESS: at sentence #70000, processed 2335266 words, keeping 194175 word types\n",
      "2017-06-05 21:01:38,923 : INFO : PROGRESS: at sentence #80000, processed 2691347 words, keeping 213170 word types\n",
      "2017-06-05 21:01:39,019 : INFO : PROGRESS: at sentence #90000, processed 3032813 words, keeping 231152 word types\n",
      "2017-06-05 21:01:39,115 : INFO : PROGRESS: at sentence #100000, processed 3358023 words, keeping 246709 word types\n",
      "2017-06-05 21:01:39,216 : INFO : PROGRESS: at sentence #110000, processed 3676488 words, keeping 263220 word types\n",
      "2017-06-05 21:01:39,320 : INFO : PROGRESS: at sentence #120000, processed 4036415 words, keeping 280933 word types\n",
      "2017-06-05 21:01:39,432 : INFO : PROGRESS: at sentence #130000, processed 4420997 words, keeping 299972 word types\n",
      "2017-06-05 21:01:39,535 : INFO : PROGRESS: at sentence #140000, processed 4770116 words, keeping 316077 word types\n",
      "2017-06-05 21:01:39,638 : INFO : PROGRESS: at sentence #150000, processed 5118100 words, keeping 331234 word types\n",
      "2017-06-05 21:01:39,747 : INFO : PROGRESS: at sentence #160000, processed 5473468 words, keeping 345836 word types\n",
      "2017-06-05 21:01:39,868 : INFO : PROGRESS: at sentence #170000, processed 5812236 words, keeping 360275 word types\n",
      "2017-06-05 21:01:39,971 : INFO : PROGRESS: at sentence #180000, processed 6164583 words, keeping 374683 word types\n",
      "2017-06-05 21:01:40,066 : INFO : PROGRESS: at sentence #190000, processed 6483079 words, keeping 387719 word types\n",
      "2017-06-05 21:01:40,164 : INFO : PROGRESS: at sentence #200000, processed 6815245 words, keeping 401395 word types\n",
      "2017-06-05 21:01:40,265 : INFO : PROGRESS: at sentence #210000, processed 7146913 words, keeping 414885 word types\n",
      "2017-06-05 21:01:40,374 : INFO : PROGRESS: at sentence #220000, processed 7519412 words, keeping 429527 word types\n",
      "2017-06-05 21:01:40,473 : INFO : PROGRESS: at sentence #230000, processed 7853284 words, keeping 442511 word types\n",
      "2017-06-05 21:01:40,577 : INFO : PROGRESS: at sentence #240000, processed 8205247 words, keeping 455872 word types\n",
      "2017-06-05 21:01:40,645 : INFO : collected 464127 word types from a corpus of 8414103 raw words and 246175 sentences\n",
      "2017-06-05 21:01:40,646 : INFO : Loading a fresh vocabulary\n",
      "2017-06-05 21:01:40,919 : INFO : min_count=10 retains 46956 unique words (10% of original 464127, drops 417171)\n",
      "2017-06-05 21:01:40,920 : INFO : min_count=10 leaves 7682424 word corpus (91% of original 8414103, drops 731679)\n",
      "2017-06-05 21:01:41,044 : INFO : deleting the raw counts dictionary of 464127 items\n",
      "2017-06-05 21:01:41,055 : INFO : sample=0.001 downsamples 20 most-common words\n",
      "2017-06-05 21:01:41,056 : INFO : downsampling leaves estimated 7521808 word corpus (97.9% of prior 7682424)\n",
      "2017-06-05 21:01:41,057 : INFO : estimated required memory for 46956 words and 200 dimensions: 98607600 bytes\n",
      "2017-06-05 21:01:41,242 : INFO : resetting layer weights\n",
      "2017-06-05 21:01:41,883 : INFO : training model with 4 workers on 46956 vocabulary and 200 features, using sg=1 hs=0 sample=0.001 negative=5 window=4\n",
      "2017-06-05 21:01:42,895 : INFO : PROGRESS: at 1.07% examples, 389918 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:43,915 : INFO : PROGRESS: at 2.18% examples, 391615 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:44,923 : INFO : PROGRESS: at 3.32% examples, 399266 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:01:45,936 : INFO : PROGRESS: at 4.38% examples, 398411 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:01:46,937 : INFO : PROGRESS: at 5.51% examples, 400711 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:47,978 : INFO : PROGRESS: at 6.56% examples, 399618 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:01:48,985 : INFO : PROGRESS: at 7.71% examples, 401919 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:49,997 : INFO : PROGRESS: at 8.84% examples, 401208 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:51,004 : INFO : PROGRESS: at 9.86% examples, 401863 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:52,035 : INFO : PROGRESS: at 10.89% examples, 401207 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:01:53,047 : INFO : PROGRESS: at 11.93% examples, 401540 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:54,064 : INFO : PROGRESS: at 12.99% examples, 401705 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:55,089 : INFO : PROGRESS: at 14.08% examples, 401493 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:56,095 : INFO : PROGRESS: at 15.16% examples, 401262 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:01:57,096 : INFO : PROGRESS: at 16.22% examples, 400063 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:58,102 : INFO : PROGRESS: at 17.35% examples, 401015 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:01:59,121 : INFO : PROGRESS: at 18.35% examples, 400503 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:00,128 : INFO : PROGRESS: at 19.46% examples, 401340 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:01,136 : INFO : PROGRESS: at 20.56% examples, 401159 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:02,158 : INFO : PROGRESS: at 21.67% examples, 401611 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:03,167 : INFO : PROGRESS: at 22.78% examples, 400972 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:04,175 : INFO : PROGRESS: at 23.80% examples, 399998 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:05,187 : INFO : PROGRESS: at 24.87% examples, 399843 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:06,200 : INFO : PROGRESS: at 25.86% examples, 398204 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:07,230 : INFO : PROGRESS: at 26.86% examples, 397515 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:08,249 : INFO : PROGRESS: at 27.87% examples, 395670 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:09,271 : INFO : PROGRESS: at 28.87% examples, 393901 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:10,296 : INFO : PROGRESS: at 29.79% examples, 392534 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:11,297 : INFO : PROGRESS: at 30.73% examples, 392137 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:12,306 : INFO : PROGRESS: at 31.73% examples, 392007 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:13,336 : INFO : PROGRESS: at 32.77% examples, 391940 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:14,345 : INFO : PROGRESS: at 33.87% examples, 392647 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:15,368 : INFO : PROGRESS: at 34.90% examples, 392350 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:16,396 : INFO : PROGRESS: at 35.92% examples, 391256 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:17,397 : INFO : PROGRESS: at 37.01% examples, 391229 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:18,402 : INFO : PROGRESS: at 38.02% examples, 391668 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:19,412 : INFO : PROGRESS: at 39.09% examples, 391782 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:20,424 : INFO : PROGRESS: at 40.15% examples, 391862 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:21,468 : INFO : PROGRESS: at 41.32% examples, 392332 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:22,476 : INFO : PROGRESS: at 42.50% examples, 392896 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:23,483 : INFO : PROGRESS: at 43.58% examples, 393013 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:24,485 : INFO : PROGRESS: at 44.69% examples, 393603 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:25,498 : INFO : PROGRESS: at 45.77% examples, 393662 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:26,505 : INFO : PROGRESS: at 46.84% examples, 394170 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:27,519 : INFO : PROGRESS: at 47.96% examples, 394212 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:28,520 : INFO : PROGRESS: at 49.11% examples, 394512 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:29,567 : INFO : PROGRESS: at 50.08% examples, 394442 words/s, in_qsize 8, out_qsize 1\n",
      "2017-06-05 21:02:30,577 : INFO : PROGRESS: at 51.10% examples, 394264 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:31,621 : INFO : PROGRESS: at 52.10% examples, 393877 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:32,623 : INFO : PROGRESS: at 53.02% examples, 393135 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:33,625 : INFO : PROGRESS: at 53.96% examples, 392405 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:34,631 : INFO : PROGRESS: at 54.90% examples, 391673 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:35,689 : INFO : PROGRESS: at 55.99% examples, 391252 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:36,689 : INFO : PROGRESS: at 57.01% examples, 390767 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:37,752 : INFO : PROGRESS: at 57.96% examples, 390173 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:38,753 : INFO : PROGRESS: at 59.00% examples, 390180 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:39,762 : INFO : PROGRESS: at 60.01% examples, 389976 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:40,777 : INFO : PROGRESS: at 61.01% examples, 389461 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:41,788 : INFO : PROGRESS: at 62.08% examples, 389436 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:42,795 : INFO : PROGRESS: at 63.24% examples, 389847 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:43,833 : INFO : PROGRESS: at 64.29% examples, 389789 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:44,838 : INFO : PROGRESS: at 65.44% examples, 390223 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:45,854 : INFO : PROGRESS: at 66.48% examples, 390307 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:46,862 : INFO : PROGRESS: at 67.62% examples, 390697 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:47,868 : INFO : PROGRESS: at 68.73% examples, 390678 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:48,870 : INFO : PROGRESS: at 69.72% examples, 390545 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:49,876 : INFO : PROGRESS: at 70.65% examples, 390509 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:50,886 : INFO : PROGRESS: at 71.66% examples, 390465 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:51,887 : INFO : PROGRESS: at 72.66% examples, 390369 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:52,895 : INFO : PROGRESS: at 73.68% examples, 390339 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:53,913 : INFO : PROGRESS: at 74.73% examples, 390389 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:02:54,933 : INFO : PROGRESS: at 75.85% examples, 390427 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:55,957 : INFO : PROGRESS: at 76.90% examples, 390075 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:56,961 : INFO : PROGRESS: at 77.86% examples, 390060 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:57,971 : INFO : PROGRESS: at 78.93% examples, 390141 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:58,989 : INFO : PROGRESS: at 79.99% examples, 390177 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:02:59,998 : INFO : PROGRESS: at 81.10% examples, 390384 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:00,998 : INFO : PROGRESS: at 82.19% examples, 390404 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:02,020 : INFO : PROGRESS: at 83.30% examples, 390531 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:03,032 : INFO : PROGRESS: at 84.41% examples, 390825 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:04,067 : INFO : PROGRESS: at 85.51% examples, 390780 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:05,069 : INFO : PROGRESS: at 86.54% examples, 390902 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:06,070 : INFO : PROGRESS: at 87.64% examples, 391018 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:03:07,090 : INFO : PROGRESS: at 88.68% examples, 390619 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:08,119 : INFO : PROGRESS: at 89.72% examples, 390603 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:09,141 : INFO : PROGRESS: at 90.65% examples, 390497 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:10,159 : INFO : PROGRESS: at 91.66% examples, 390430 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:11,171 : INFO : PROGRESS: at 92.68% examples, 390404 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:12,228 : INFO : PROGRESS: at 93.81% examples, 390567 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:13,232 : INFO : PROGRESS: at 94.88% examples, 390756 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:03:14,234 : INFO : PROGRESS: at 95.92% examples, 390570 words/s, in_qsize 8, out_qsize 0\n",
      "2017-06-05 21:03:15,236 : INFO : PROGRESS: at 97.03% examples, 390664 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:16,246 : INFO : PROGRESS: at 97.98% examples, 390535 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:17,262 : INFO : PROGRESS: at 99.09% examples, 390751 words/s, in_qsize 7, out_qsize 0\n",
      "2017-06-05 21:03:18,040 : INFO : worker thread finished; awaiting finish of 3 more threads\n",
      "2017-06-05 21:03:18,072 : INFO : worker thread finished; awaiting finish of 2 more threads\n",
      "2017-06-05 21:03:18,090 : INFO : worker thread finished; awaiting finish of 1 more threads\n",
      "2017-06-05 21:03:18,097 : INFO : worker thread finished; awaiting finish of 0 more threads\n",
      "2017-06-05 21:03:18,098 : INFO : training on 42070515 raw words (37609839 effective words) took 96.2s, 390914 effective words/s\n"
     ]
    }
   ],
   "source": [
    "word_two_vec = models.Word2Vec(texts, size=200, window=4, min_count=10, workers=4, sg=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "gensim.models.word2vec.Word2Vec"
      ]
     },
     "execution_count": 123,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "type(word_two_vec)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.642901638534\n"
     ]
    }
   ],
   "source": [
    "print(word_two_vec.similarity('elon','tesla'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's try some pair-wise comparisons! "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "nice and linkedin: 0.20239235462425015\n",
      "helpful and linkedin: 0.3876046858402107\n",
      "awesome and linkedin: 0.3498346593234357\n",
      "jobs and linkedin: 0.29799666052627405\n",
      "good and linkedin: 0.24090598433809327\n",
      "bad and linkedin: 0.23487223532511387\n",
      "spam and linkedin: 0.5632705948911862\n",
      "advertising and linkedin: 0.3759218337755161\n",
      "ads and linkedin: 0.3505997308576099\n",
      "ad and linkedin: 0.35901339131943133\n",
      "spam and linkedin: 0.5632705948911862\n",
      "recruit and linkedin: 0.5795595281831671\n",
      "job and linkedin: 0.35215434476566415\n",
      "network and linkedin: 0.25516178153332936\n",
      "founder and linkedin: 0.49163205187247555\n"
     ]
    }
   ],
   "source": [
    "words_with_linkedin = ['nice', \"helpful\", \"awesome\", \"jobs\", 'good', 'bad', 'spam', 'advertising', 'ads', \"ad\", 'spam', \"recruit\", 'job', 'network', 'founder']\n",
    "\n",
    "for word in words_with_linkedin:\n",
    "    similarity = word_two_vec.similarity('linkedin', word)\n",
    "    print(\"{0} and linkedin: {1}\".format(word, similarity))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "nice and thiel: 0.19632102680414576\n",
      "good and thiel: 0.24903499217676953\n",
      "bad and thiel: 0.2511811258988722\n",
      "job and thiel: 0.20755877484129787\n",
      "network and thiel: 0.13711916392128082\n",
      "founder and thiel: 0.5909898758604776\n",
      "fund and thiel: 0.32248024521926766\n",
      "college and thiel: 0.3060743308021644\n"
     ]
    }
   ],
   "source": [
    "words_with_thiel = ['nice', 'good', 'bad', 'job', 'network', 'founder', 'fund', 'college']\n",
    "\n",
    "for word in words_with_thiel:\n",
    "    similarity = word_two_vec.similarity('thiel', word)\n",
    "    print(\"{0} and thiel: {1}\".format(word, similarity))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cat and tesla: 0.21505069992285658\n",
      "cars and tesla: 0.5988400836137333\n",
      "mars and tesla: 0.3775405114799705\n",
      "musk and tesla: 0.6253844894378401\n",
      "car and tesla: 0.5526767085173219\n",
      "self and tesla: 0.38478363577643027\n",
      "three and tesla: 0.03963286818319984\n",
      "3 and tesla: 0.23641768562017917\n",
      "model and tesla: 0.3328042240316319\n",
      "spacex and tesla: 0.6253956259353688\n",
      "elon and tesla: 0.6429016385343611\n",
      "electric and tesla: 0.601355724488056\n",
      "vehicle and tesla: 0.5609685808100657\n"
     ]
    }
   ],
   "source": [
    "words_with_tesla = ['cat', 'cars', \"mars\", \"musk\", 'car', 'self', 'three', '3', 'model', 'spacex', 'elon', 'electric', 'vehicle']\n",
    "\n",
    "for word in words_with_tesla:\n",
    "    similarity = word_two_vec.similarity('tesla', word)\n",
    "    print(\"{0} and tesla: {1}\".format(word, similarity))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "science and data: 0.1616682181523955\n",
      "machine and data: 0.30986394282397633\n",
      "ai and data: 0.21883042231688965\n",
      "deep and data: 0.17386263496158044\n",
      "neural and data: 0.25426244585394137\n",
      "network and data: 0.2575073539589451\n",
      "learning and data: 0.21901204020162834\n",
      "model and data: 0.23399699345593594\n",
      "warehouse and data: 0.31522597025306526\n"
     ]
    }
   ],
   "source": [
    "words_with_data = ['science', 'machine', \"ai\", \"deep\", \"neural\", \"network\", \"learning\", \"model\", \"warehouse\"]\n",
    "\n",
    "for word in words_with_data:\n",
    "    similarity = word_two_vec.similarity('data', word)\n",
    "    print(\"{0} and data: {1}\".format(word, similarity))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cars and uber: 0.4613491189487633\n",
      "drivers and uber: 0.5660130195233564\n",
      "driver and uber: 0.5198409876794874\n",
      "taxi and uber: 0.6967426203582391\n",
      "car and uber: 0.4399795883843042\n",
      "self and uber: 0.344473022757424\n",
      "travis and uber: 0.4635756352604215\n",
      "lyft and uber: 0.7824413043944751\n",
      "share and uber: 0.19054697015758365\n"
     ]
    }
   ],
   "source": [
    "words_with_uber = ['cars', \"drivers\", \"driver\", \"taxi\", 'car', 'self', 'travis', 'lyft', 'share']\n",
    "\n",
    "for word in words_with_uber:\n",
    "    similarity = word_two_vec.similarity('uber', word)\n",
    "    print(\"{0} and uber: {1}\".format(word, similarity))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cars and apple: 0.27135330351255105\n",
      "self and apple: 0.22822908757019206\n",
      "steve and apple: 0.411437767835246\n",
      "jobs and apple: 0.16086889934329113\n",
      "ios and apple: 0.5704511588097121\n",
      "android and apple: 0.5176579568782066\n",
      "google and apple: 0.44813559423265015\n",
      "linux and apple: 0.3587924070906193\n",
      "os and apple: 0.4633175147944112\n",
      "mac and apple: 0.5665887571861825\n"
     ]
    }
   ],
   "source": [
    "words_with_apple = ['cars', 'self', 'steve', 'jobs', 'ios', 'android', 'google', 'linux', 'os', 'mac']\n",
    "\n",
    "for word in words_with_apple:\n",
    "    similarity = word_two_vec.similarity('apple', word)\n",
    "    print(\"{0} and apple: {1}\".format(word, similarity))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With these pair-wise comparisions, we get a better understanding of "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3.0
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}