Credibility Microscope: Relating Web Page Credibility Evaluations to Their Textual Content

The popularization of the Internet made it a primary information source for many people. Unfortunately quality of information available on the Internet varies. Therefore, evaluation of credibility of web page content, especially while making important decisions like those concerning health care, medical information, and large purchases, is crucial, but users often lack a necessary knowledge. The main goal of the paper is to study to what extent the textual content of a webpage determines its credibility evaluations. This goal is achieved by an experiment in which we ask respondents to rate webpage credibility as well as credibility and importance of each statement from this site. We formulate a number of hypotheses about the nature of the dependence between webpage and statements credibility and we test these hypotheses on the data obtained from experiment. The evaluation of those hypotheses is essential for design of classifier that will aggregate statement credibility into webpage credibility score.