Fake News Detection in Social Networks Using Machine Learning and Trust

Fake news propagation is a major challenge for Online Social Networks (OSN) security, which is not yet resolved. Fake news propagates because of several reasons, one of which is non-trustworthy users. Non-trustworthy users are those who spread misleading information either for malicious intentions or innocently as they lack social media awareness. As a result, they expose their sub networks to false or inaccurate information. In our previous research we have devised a comprehensive Trust-based model that can handle this problem from the user Trust aspect. The model involvesAccess Control for the direct circle of friends and Flow Control for the friends' networks. In this paper we use this model and extend it for the purpose of preventing Fake News. We analyze user's activity in the network (posts, shares, etc.) to learn their contexts. Using Machine Learning methods on data items that are fake or misleading, we detect suspicious users. This addition facilitates a much more accurate mapping of OSN users and their data which enables the identification of the Fake News propagation source. The extended model can be used to create a strong and reliable data infrastructure for OSN.