A multi-criteria decision making approach for the assessment of information credibility in social media

In Social Media, large amounts of User Generated Content (UGC) generally diffuse without any form of trusted external control. In this context, the risk of running into misinformation is not negligible. For this reason, assessing the credibility of both information and its sources in Social Media platforms constitutes nowadays a fundamental issue for users. In the last years, several approaches have been proposed to address this issue. Most of them employ machine learning techniques to classify information and misinformation. Other approaches exploit multiple kinds of relationships connecting entities in Social Media applications, focusing on credibility and trust propagation. Unlike previous approaches, in this paper we propose a model-driven approach based on Multi-Criteria Decision Making (MCDM) and quantifier guided aggregation. An overall credibility estimate for each piece of information is obtained based on multiple criteria connected to both UGC and users generating it. The proposed model is evaluated in the context of opinion spam detection in review sites, on a real-world dataset crawled from Yelp, and it is compared with well-known supervised machine learning techniques.