*Theme settings will be saved for
your next vist
Created by: Dylan Dalton, Yesse Quezada, Ray Martin, Duy Nguyen, & Brian Dominguez
This project is to develop a web based app or tool that leverages a natural language processing (NLP) model to evaluate the bias of online news articles shared through social media sites such as Reddit
for power point and more infomation click Download
More and more people are getting their news from social media exposing them to a wider variety of journalistic institutions. Journalism will always contain bias, the goal of this project is to warn potential end users about articles that may contain strongly biased content.
Input desired URL to a news article in the app. The app will scrape the article’s text. The text is cleaned and embedded using Doc2Vec from Gensim. The vectorized text is passed into a neural net which is then scored on an economic left-right scale, and authoritarian-libertarian scale. The scores are then displayed on a graph.
Our data collection consists of two parts and has been the bulk of our work so far. We use the Reddit API to pull comments from ideologically aligned subreddits. So far, we have collected about 50,000 comments from reddit, but we are always pulling more. The other part of data collection is scraping news articles. We have implemented a bot that goes through the posts we have pulled from reddit and extracts the articles text from each of the new article links that have been shared.