Architecture of a fake news detection system combining digital watermarking, signal processing, and machine learning

Abstract

In today’s world, the ease of creation and distribution of fake news is becoming an increasing threat for individuals, companies, and institutions alike. Content spread over the Internet is able to create an “alternative” reality and false accusations cannot be easily removed by later issued apologies as it typically takes several years to unpick the labels pinned on by spreading disinformation. Currently, the main facilitators of fake news distribution are social media networks, where a large volume of digital media content is generated and exchanged every day. In this “flood” of information, it is quite effortless to manipulate the content to impact its consumers. That is why developing effective coun- termeasures is of prime importance. Considering the above, in this paper, we propose and describe an architecture of the fake news detection system that is being developed within an ongoing Detection of fake newS on SocIal MedIa pLAtfoRms (DISSIMILAR) project. It is designed for the protection of digital media content, i.e., images, video, and audio, and to fulfill its goals, it combines digital watermarking, signal processing, and machine learning techniques.