Automated Techniques for Identifying Claims and Assisting Fact Checkers
Stefano Agresti, Mark Carman · 2023
A concerning issue of our age is the spread of misinformation online. The dissemination of false, misleading, or biased information can adversely affect society and can threaten democracy by precluding healthy, fact-based political discourse. The circulation of false or distorted news can sow distrust and fear among the public, culminating in the propagation of conspiracy theories. Not only can this exacerbate political partisanship, but also it can make it difficult to enforce unpopular, yet necessary, legislation as seen during the COVID-19 pandemic. While it would be naïve to place all the blame for the spread of such misinformation on social media platforms, it is undeniable that social networks have allowed fake news to prosper as never before. Numerous studies have investigated how best to fight this phenomenon, oftentimes exploiting powerful Artificial Intelligence techniques. Yet, they suffer from a limitation when dealing with such an elusive problem in that they assume a simple dichotomy between real and fake news. In this chapter, we propose a new finer-grained taxonomy for online news content that goes well beyond this binary distinction. We investigate the feasibility of categorizing texts according to this new classification scheme, using datasets extracted from Reddit, and a crowdsourcing-based evaluation. Then based on these classifiers, we prototype a system for assisting journalists with their fact-checking activities, helping them to identify claims in political transcripts and retrieve passages from news articles that may provide evidence supporting or refuting those claims.