jcwhuang/naive-twitter-tagger

Naive Bayes Classifier that "classifies" tweets by predicting "relevant" hashtags (originally Intro to Machine Learning final project)

naive-twitter-tagger

Originally created as a final project for Intro to Machine Learning by Matthew Hauser and Jade Huang.

Description

naive-twitter-tagger is a Naive Bayes classifier that "auto-generates" relevant Twitter hashtags. Relevancy is determined based on frequency of a word and hashtag occurring together as well as frequency of a hashtag. This could potentially be useful to help generate descriptive tags for users to increase their visibility and searchability.

Our training data consists of a mix of precollected corpuses of tweets as well as randomly collected tweets using two libraries of Python scripts and the Twitter API.

Precompiled corpuses

• Sentiment140 Corpus
• Twitter Political Corpus

Scripts used to generate corpuses

• Python Twitter
• [twitter-stream-downloader](https://github.com/mdredze/twitter stream downloader)