Fake News Filtering: Semantic Approaches

Abstract

In 2016, the attention to the fake news phenomenon drastically increased. Mobile devices such as cellular phones and sources of information such as social networks are instruments that enable individuals to receive news, publish posts, communicate with peers, watch videos, listen to music, etc. In today’s highly mobile society, this is a current trend. The uncontrolled freedom and simplicity in publications on the Internet result in overwhelming users receiving news that are fake and hoaxes. Detecting and filtering such information is a challenging problem. This paper discusses different approaches to combat fake news. They are used to a) determine text features utilizing linguistic natural language processing methods (it is necessary to create a profile of the text document), b) detect spam bots in social networks to isolate those using machine-learning methods (it is crucial to reduce the number of analyzed documents), and c) confirm the facts in online documents by applying techniques used in search engines (it is very much important to select trusted documents). A system combining these mechanisms may demonstrate a high level of accuracy in filtering fake news.