Credibility Assessment in Microblogs

Abstract:

Microblog platforms allow users to post publicly viewable short messages. Unfortunately, not all posted information may be credible and these platforms have become a prime target for the spread of fake news and misinformation when some major event occurs. Despite advances in tools and techniques for the detection and verification of fake news, assessing the credibility of information remains a challenge.

We observe that a micropost typically contain multiple pieces of information, each of which may have different level of credibility. We first introduce the notion of a claim based on subject and predicate terms, and draw upon work done in open information extraction to extract from microposts, tuples that comprises of subjects and their predicate. Then we cluster these tuples to identify claims such that each claim refers to only one aspect of an event. Next, we propose an interactive framework called iFACT for assessing the credibility of claims. The proposed framework collects independent evidence from web search results and utilizes features from the search results to estimate the likelihood of a claim being credible, not credible or inconclusive. In addition to sourcing for external independent sources of evidence in the form of WSRs, we also identify the dependencies among claims, and use these dependencies to adjust the likelihood estimates of a claim being credible, not credible or inconclusive. Finally, we examine how the framework can be expanded for time sensitive claims, where the credibility of claims depends not just on the content, but also on the time period that the claim is purported to be valid for. We generate alternate claims and consider relationships between these alternate claims and the target claim, in order to perform a joint credibility assessment for all the claims.