A Survey Of Automatic Fact Verification Research

With the evolution of social media technology, the mode of information exchange has transitioned from a singular text format to a multifaceted blend of text, images, audio, and video. This transformation has given rise to the proliferation of multimodal false information, sparking widespread concern across diverse sectors of society. This paper takes a fact-checking perspective, offering a comprehensive review of pertinent research in the field. It commences by introducing prevalent datasets used in contemporary fact-checking. Subsequently, a detailed analysis of fact-checking methods and multimodal detection techniques, centered around the FEVER dataset, is presented, accompanied by comparative insights into various approaches. Finally, the paper delves into the anticipated challenges faced by fact-checking methods in the future.