ABSTRACT
Social media websites and applications contain data spanning sponsored contents, and shared information. Accessible information-sharing in these platforms makes the propagation of misinformation or fake news faster. In this study, the authors created a Bayes reliability classifier of news articles using topic generation and sentiment analysis. News articles were scraped from Facebook and selected news sites’ RSS feeds, and were pre-processed to populate topics and rate sentiments. Furthermore, articles were fed in a classifier that outputs value for each article based on its topics and sentiment. The generated values for each article were tested against a threshold the researchers flagged as standard reliability value. From the results, it can be concluded that news articles from popular local news companies are reliable as they yielded positive statistical mean and positive-neutral statistical mode for sentiment rating. Moreover, news articles from a certain Facebook account known for its strong political statements were classified unreliable.
