Interactive Reputation Systems: How to Cope with Malicious Behavior in Feedback Mechanisms

Abstract

Early reputation systems use simple computation metrics that can easily be manipulated by malicious actors. Advanced computation models that mitigate their weaknesses, however, are non-transparent to the end-users thus lowering their understandability and the users’ trust towards the reputation system. The paper proposes the concept of interactive reputation systems that combine the cognitive capabilities of the user with the advantages of robust metrics while preserving the system’s transparency. Results of the evaluation show that interactive reputation systems increase both the users’ detection ability (robustness) and understanding of malicious behavior while avoiding trade-offs in usability.