Sentiment Analysis and NLP models for Identifying Biases of Online News Stations




Cox, Grace
Acharya, Anuska

Journal Title

Journal ISSN

Volume Title



This work attempts to identify potential reporting bias surrounding recent controversial decisions for articles published from August 2019 to October 2021 within four major news organizations: FOX, CNN, NBC, and NPR. This potential bias is determined by conducting a Sentiment Analysis using NLTK’s (Natural Language Tool Kit) VADER (Valence Aware Dictionary and sEntiment Reasoner) Sentiment Intensity Analyzer. Copious amounts of literature have been published regarding sentiment analysis and bias identification of news articles, though none employ VADER. The team determines the overall sentiment for an article using the Polarity Compound calculated by the Sentiment Intensity Analyzer, which then corresponds to a political tone indicated within the verbiage and context of the article. Upon completion of the analysis, it was found that CNN, NBC, and NPR tend to have the most negative sentiment surrounding this topic, while FOX tends to be more neutral though still on the positive side. This translates to the surprising identification of a slightly democrat tone for articles published by FOX, and a more republican tone for those articles published by NPR, CNN, and NBC.



Machine learning, Sentiment analysis, Natural language processing, News