overview.md

Relevant Information/Links

• The Skip-gram Model
  • Backbone and base inspiration for this project
• Word Embeddings
  • Math
• Object Serialization
  • This makes everything cheaper
• BeautifulSoup Documentation
  • Python package for web scraping
• Synonym Clustering with KMeans
  • Due to computational restrictions, our implementation for synonym clustering relied on a cascading-style approach, loosely inspired by binary search.
• Skip-thoughts for Sentiment Analysis
  • For future improvements, we can use this to detect sentiment bias in article content, and perhaps make modifications for political bias detection.

Project Overview

• Note: A little outdated, but still generally correct

Important Requirements (continously updated)

• Python 2.7
• TensorFlow 1.0.1+
• scikit-learn
• matplotlib
• scipy
• numpy
• BeautifulSoup
• Django
• TBD

Frontend Components

• Basic UI allowing user login with Facebook
• API to dictate communication between frontend and backend controllers
• ReactJS

Backend Components

• Skip-gram model to vectorize and train on semantics of words
• Django backend with PostgreSQL DB
• Webcrawler to find and classify new data using the model, then feed the user relevant info