CHARACTERIZATION OF FAKE NEWS BASED ON SUBJECTIVITY LEXICONS

摘要

While many works investigate spread patterns of fake news in social networks, we focuson the textual content. Instead of relying on syntactic representations of documents(aka Bag of Words) as many works do, we seek more robust representations that maybetter differentiate fake from legitimate news. We propose to consider the subjectivityof news under the assumption that the subjectivity levels of legitimate and fake newsare significantly different. For computing the subjectivity level of news, we rely on aset subjectivity lexicons for both Brazilian Portuguese and English languages. We thenbuild subjectivity feature vectors for each news article by calculating the Word Mover'sDistance (WMD) between the news and these lexicons considering the embedding thenews words lie in, in order to analyze and classify the documents. The results demon-strate that our method is robust, especially in scenarios where training and test domainsare different.