Computational Study on Strategyproofness of Resource Matching in Crowdsourced Manufacturing

抄録

The need for a sustainable society has grown rapidly. This trend requires new production system concepts following an era of mass customization. As one of these new concepts, “crowdsourced manufacturing” has attracted noticeable attention. In such systems, each participant shares their manufacturing resources for ecosystem co-prosperity, providing new value for the next society. To realize such a concept, it is important to (1) match resource requests and resource offers so as to achieve high efficiency, and (2) induce participants to act in a fair way. Previously, some studies showed production efficiency improvements. Nevertheless, relatively few studies have been conducted on induction mechanisms. The purpose of this study is to develop induction mechanisms for participants. Concerning induction mechanisms, we focus on two viewpoints: (a) matching stability, and (b) “strategyproofness.” These viewpoints are well-known concepts in the market design research field. We previously proposed a resource matching stability analysis method and mechanism for inducing participants to accept matching plans. Formally, a matching method is “strategyproof” when it is a dominant strategy for all participants to submit their true information. However, it is hard to satisfy this condition. Practically, it would be useful to evaluate the strength of an induction, even if the matching method is not strategyproof. In this study, we propose indices for showing the strength of induction (“strength of strategyproofness”). Subsequently, we evaluate matching methods, and show that participants will state false information to maximize their profit in a system with resource matching methods for the profit maximization of the entire system. As the resource providers, they can obtain greater profit by submitting false information regarding resource usage fees. Then, the profits of the resource requesters are unfairly impaired. Furthermore, we propose a new resource matching method, inspired from the “nucleolus” concept in cooperative game theory. The proposed method reduces the maximum dissatisfaction (i.e., profit loss) of resource requesters and resource providers, based on profit sharing. The computational results show that the proposed method induces participants to submit true information, while maintaining high production efficiency.