TrustSets - Using Trust to Detect Deceitful Agents in a Distributed Information Collecting System

This paper presents a study on how to improve a distributed information collecting system in which information is collected by a multi-agent system constituted by communicating agents, assuming the hypothesis that some agents of this system can deliberately (liar agents) or in good faith (defective agents) produce or communicate incorrect information. To ensure the coherence of the information system under these constraints, we aim to gradually limit the impact of the incoming perturbations. To reach this goal, we propose that each agent develops its own communication strategy from a TrustSet it builds using information collected by itself and information received from agents it communicates with.