Człowiek i maszyna w walce z fake news: porównanie analizy cech lingwistycznych dezinformacji dokonanej przez człowieka i sztuczną inteligencję - narzędzie uczenia maszynowego

Abstract

The term ‘fake news’ is now firmly established in public discourse and collective consciousness; Internet disinformation is a serious problem which is capable of shaking the foundations of democracy. One method of detecting fake news is to use machine learning techniques; ideally, these tools should be ‘explainable’. The aim of this paper is to present a set of linguistic features indicative of fabrication of news, to perform a human analysis of these features, to determine the veracity messages by means of artificial intelligence – a machine learning tool, and to test whether a human researcher and the machine learning algorithm recognize fake news by paying attention to the same linguistic features of the messages.