bayesianfakenewsclassifier

Twitter Fake News Classifier (Bayes Theorem) The use of the Internet as a means of accessing news content has become part of the daily lives of most of its users. In particular, social networks, driven by the technological advancement of mobile Internet and mobile devices, have become major sources of information. With the popularity of such social networks, the spread of fake news has increased considerably in recent years. Such news are called fake news, which have a structure similar to a real news, and are created in order to gain advantages in various sectors, such as politics, sports, among others. One of the main social networks today is Twitter, which has millions of active users and allows the collection of information from its database through libraries specific for this purpose. From the data collected from this social network, it is possible to analyze objects of these data through various algorithms in the computer literature, among which is the Naive Bayes classifier, commonly used in the classification of textual documents. In this sense, this project aims to present a classifier capable of detecting fake news, having as input a database composed of news collected on the social network Twitter. Initially, we collected and stored the tweets that will be analyzed and, from then on, the Naive Bayes classifier was used in the experiment that determined the veracity of the information, such experiment obtained 85% maximum accuracy. With the results of this classification step, empirical analyzes were performed in order to verify if these results approximate the labeled value. Thus, quantitative and qualitative analyzes were performed, in which it was possible to infer the correlation of the results obtained with reality.