Skip to content

donotabide/NLPemo

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

41 Commits
 
 
 
 

Repository files navigation

Information Retrieval/Extraction in Emotional-Solution Entries

This project intends to perform information extraction of emotional entries and matching them with potential solution paragraphs. Our goal is to perform an assessment of users emotional states, giving them a platform to post, relate to others and receive potential solutions anonymously.

Overview

Using past emotional entries the program suggests a potential solution for a user’s current emotional state. The program assigns emotional “weights” and draws from a database of potential solution based on these emotional weightages.

Data

Emotional Word Bank Text.txt document containing a list of words and their associated emotions each entry in a new line afraid \t fear \t sadness \n

Emotional Tag’s

  1. Happiness 2) Sadness 3) Fear/Surprise 4) Anger/Disgust

Gathering Data

Entries from Students about their current “status” on NYU Secrets and other colleges Secret pages - A Facebook page that posts thoughts from anonymous university students. These entries are used to create our emotional word bank. Key words from each entry are manually selected and emotional tags assigned. Words are selected manually due to the use of “slang” in common speech e.g “Going nuts”

Other potential data sources

Extract - Twitter API and other web scraping API

Potential Solutions

Text.txt document containing a list of potential solutions each entry in a new line. Each potential solution has emotional weights attached. E.g Slow down your breathing and read a book \t happiness 0 sadness 30 fear/surprise 70 anger/disgust 0 Potential solutions taken from the comments section of the facebook posts. We also aim to create our own potential solutions.

Data Sources:

*“Secrets” pages on Twitter and Facebook * Entries as paragraph problem * Use comments as possible solutions

  • Medical Subject Headings thesaurus (MeSH) to promote sentiment analysis of whether the entry is emotional or not

Input and Output Data Format

Input

User enters an input in the command line console. e.g “I spilled my coffee on my shirt, almost lost my phone, had a pop quiz that went terribly but was able to pet two adorable golden retriever puppies on the subway today”. Output The program outputs a suggested solution or comment E.g “Take a deep breath. You’ll be fine”

Process

SenPy, NLTK using Sentiment Analysis and Information retrieval, give weights to pre-defined emotions and solutions.

Alternative 1: The Algorithm (Input as problem paragraph)

  • Summarization of paragraph
  • Detect “key words”
  • Match “key words” with the emotional word bank
  • Assign emotional score based on 4 categories of emotional words
  • Based on the score match 2 - 5 closest solutions
  • Provide one of the possible best solution to the users

Alternative 2: as Queries (input)

  • Summarize problem paragraph
  • Sentiment analysis: is it an emotional entry or not?
  • Make it into a query vector
  • Run search in database to see correlation in database
  • Number of documents with the vector words
  • Run vector through database to see which solution paragraph has best match
  • Return best match

Responsibilities

Arjun Madgavkar:

  • Data scraping & formatting
  • Assigning weights based on input and emotional bank
  • iPhone application (designs and development, if we want to create one)

Han Zhang:

  • Building emotional words bank
  • Matching input to an existing bank of solutions vectors

Jaeduk Choi:

  • assigning weights based on input and emotional bank
  • building emotional words bank

Nicholas Ang:

  • Data scraping & formatting
  • Building solutions word bank and assigning attributes to solutions

Ricardo Nunes:

  • Data scraping and formating
  • Emotional words bank
  • Data processing using vectors

Possible Paid Algorithms

URL sources:

About

No description, website, or topics provided.

Resources

Stars

1 star

Watchers

3 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages