Identifying Fake News using Emotion Analysis
Brady Gilleran · Journal of the Arkansas Academy of Science · 2019
This paper presents research applying Emotional Analysis to “Fake News” and “Real News” articles to investigate whether or not there is a difference in the emotion used in these two types of news articles. The paper reports on a dataset for Fake and Real News that we created, and the natural language processing techniques employed to process the collected text. We use a lexicon that includes predefined words for eight emotions (anger, anticipation, disgust, fear, surprise, sadness, joy, trust) to measure the emotional impact in each of these eight dimensions. The results of the emotion analysis are used as features for machine learning algorithms contained in the Weka package to train a classifier. This classifier is then used to analyze a new document to predict/classify it to be “Fake” or “Real” News.