Fake consensus in online news comment section

Abstract

Manipulation of public opinion over the Internet has emerged as a critical threat to our society across the globe. In this dissertation, we coin the term fake consensus to refer to a type of information manipulation focusing on amplifying or undermining the popularity of opinions rather than the truth or falsehood of content. In online news comment sections, such fake consensus is manufactured directly by massively clicking upvotes and downvotes to comments with coordinated efforts and indirectly by uploading duplicate comments multiple times using multiple accounts. This dissertation presents quantitative studies about the fake consensus problem in comment sections in giant South Korean news portals. We have collected and analyzed the entire articles and comments from three giant South Korean news portal sites in the last decade. Also, we examined news sharing and targeting behavior from partisan online communities. Using these datasets, we developed novel measures to identify and quantify coordinated manipulation of upvotes and downvotes and measured its prevalence in Naver News comment sections. Also, by analyzing duplicate comments and the uploader networks, we showed that such manipulative behavior becomes more participatory and less organized in recent years. These quantitative studies provide records about the growth of South Korean news portals, the news user behavior, the history of opinion manipulation in the comment sections, and the recent trends of how such manipulation has changed into.