Machine Learning on Hacker News in JavaScript
JavaScript
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
lib
public
Jakefile
README.md
app.js
config.js
package.json
sample-new.json

README.md

HN Learning

This was a fun projec to play with machine learning in JavaScript.

It uses credulous, a Naive Bayes model written in JavaScript, to build up a model for classifying posts as interesting or uninteresting.

Installation

You need node.js and mongodb installed already.

From inside the hn-learning directory, run:

npm install
npm install -g jake
mongod & #start mongodb in the background

Using It

Run jake -T to see the Jake tasks. This will get you started rating posts.

jake load_from_hnarchive # seed the database with stuff
node lib/commandLineRater.js #begin rating posts

Once you want to classify new posts, run node lib/recommendNewPosts.js. This will pull 30 posts from the Hacker News new page and recommend ones to you that are recommended by the model.