Comparison of six aggregation strategies to compute users' trustworthiness

Abstract

The decision to grant trust in virtual societies is often an evidence based process. The evidence for such decision derives from a diverse set, where mutual relationships and contradictions might occur. This paper compares and evaluates six aggregation strategies to compute users' trustworthiness. Our evaluation performed over a large online-community, shows how a rule-based strategy based on an argumentation semantic outperforms strategies where mutual relationships among evidence are ignored.