MODELING CREDIBILITY ASSESSMENT AND EXPLANATION FOR TWEETS BASED ON SENTIMENT ANALYSIS

Credibility is one of the main issues when dealing with information obtained from Online Social Networks (OSNs). Although a significant number of prior works have addressed many issues in this topic, only a few that have worked on methods for automatic credibility measurement for OSN messages and almost none who has addressed a specific problem in explaining the credibility information. This paper proposes a new approach for modeling credibility of tweets and explaining it to users. We model tweet credibility based on other independent tweet contents that support and oppose the topic issue in question. We also consider the opinions of tweets’ followers who either confirm or deny, along with their reputations. This method is based on assumption that the more community is who agree with the claim, the more credible is the claim. Explanation of the credibility measure is based on the tweet content itself. We provide users with representative tweets that can be progressively zoomed-in for more detail explanation. To achieve this goal, all tweets are hierarchically structured and tweet representatives on each node are selected from the ones that are most similar to the centroid. Our evaluation results indicate the feasibility of the proposed methods.