Partner selection is a challenge central to dynamic peer-to-peer (P2P) supply chain networks with decentralized architecture. In this study, we examine how ratings information can be used to select the right trading partners in a P2P network. Using data collected from a bitcoin platform, we first show that the behavior patterns of fraudsters differ from those of other users. We analyze the topology of a P2P network and demonstrate that it is a small-world network and that various internal and external factors can affect the ratings. We find that it is also temporal and that effective rating indices for partner selection are also time-varying. We design a system to simulate the partner selection process and access its impacts on the effectiveness of the ratings information under different information revelation policies—all past information, three-month information, and one-month information—along with different levels of tolerance for deception. Our results provide practical insights into reputation system design in a P2P platform and have implications for partner selection in a P2P supply chain network.

合作伙伴选择是具有去中心化架构的动态点对点 (P2P) 供应链网络的核心挑战。在这项研究中，我们研究了如何使用评级信息来选择 P2P 网络中的正确贸易伙伴。使用从比特币平台收集的数据，我们首先表明欺诈者的行为模式不同于其他用户的行为模式。我们分析了 P2P 网络的拓扑结构，并证明它是一个小世界网络，各种内部和外部因素都会影响评级。我们发现它也是时间性的，并且合作伙伴选择的有效评级指数也是随时间变化的。我们设计了一个系统来模拟合作伙伴选择过程，并在不同的信息披露政策下访问其对评级信息有效性的影响——所有过去的信息，三个月的信息和一个月的信息——以及不同程度的欺骗容忍度。我们的结果为 P2P 平台中的声誉系统设计提供了实用的见解，并对 P2P 供应链网络中的合作伙伴选择产生了影响。