DRIS: Direct reciprocity based image score enhances performance in collaborate computing system

Abstract

The key issue in collaborate computing systems is to induce agents work together towards a common task. Indirect reciprocity is a widely used method to promote cooperation. Image score is a classic indirect reciprocity mechanism which provided an insight on promoting cooperation. However, intuitively, people trust their own feelings more than other’s opinions. Thus, in this paper, we introduce direct reciprocity into image score: when granting a service, a player should consider their own interaction histories in priority and consider image score if they have no interactions before. Extensive simulation results show that our proposed model can more effectively promote cooperation than classic image score under both complete and incomplete information.