Using CHI-scores to reward honest feedback from repeated interactions

Online communities increasingly rely on reputation information to foster cooperation and deter cheating. As rational agents can often benefit from misreporting their observations, explicit incentives must be created to reward honest feedback. Reputation side-payments (e.g., agents get paid for submitting feedback) can be designed to make truth-telling optimal. In this paper, we present a new sidepayment scheme adapted for settings where agents repeatedly submit feedback. We rate the feedback set of an agent, rather than individual reports. The CHI-Score of the feedback set is computed based on a Chi-square independence test that assesses the correlation between the agent's feedback and the feedback submitted by the rest of the community. The mechanism has intuitive appeal and generates significantly lower costs than existing incentive-compatible reporting mechanisms. Categories and Subject Descriptors I.2.11 [Artificial Intelligence]: Multiagent systems General Terms Algorithms, ...

• Type: Conference
• Year: 2006
• Where: ATAL
• Authors: Radu Jurca, Boi Faltings
• Keywords: Agent's Feedback, ATAL 2006, Feedback Set, Honest Feedback, Intelligent Agents