摘要： 提出基于聚类的P2P电子商务信誉评价算法。该算法根据节点行为的相似性将节点聚类成若干集合，将每个节点集合视为一个节点进行迭代信誉计算。诚实节点行为相似性较高，通常聚类为最大的节点集合，在迭代计算中占主导地位。实验结果表明，与基于社会网络和概率估计的信誉评价体系算法相比，该算法误差较小。

关键词: P2P信誉评价, 电子商务, 聚类, 群体作弊

Abstract: This paper presents an aggregation-based reputation evaluation algorithm for P2P e-commerce. The main feature of this algorithm is aggregating peers into several peers groups according to their behavior similarities, and the reputation of these groups is calculated iteratively. The group with the largest amount of peers has priority in the calculation of reputation. The honest peers will aggregate into one group because of their behavior similarity. Experimental results show that this algorithm has less variance compared with the algorithms of social networks or probabilistic estimation based reputation evaluation system.

Key words: P2P reputation evaluation, e-commerce, aggregation, malicious collectives

中图分类号: 

• TP393.08