Assessing the Credibility of Claims on the Web

Abstract

In my doctoral research, I plan to address the problem of assessing the credibility of arbitrary claims made in natural-language text - in an open-domain setting. Automatic credibility assessment is a complex task depending upon many factors. To start with, we propose three factors which can help in assessing the credibility of textual claims: (i) the reliability of the web sources talking about the claim, (ii) the language style of the articles reporting the claim and, (iii) their stance (i.e., support or refute) towards the claim. In addition, we also focus on extracting user-interpretable explanations as evidence supporting the verdict of the assessment.