Abstract: Aimming at the deficiency of traditional Web community discovery algorithm and the problem of cluster strength measure，the object of Web community nodes and edges and structure is given.A new cluster strength measure method is researched，in order to settle the problems of community optimal partitioning and subject optimization.Object function of maximal community is presented，community discovery algorithm based on cluster ranking of integrated cohesion is described，and application system is built based on existing information searching kit.The result of experiment shows that the algorithm can fast，effectively search global optimum partition of network structure.This algorithm is highly effective and valuable in practice and academic study.

Key words: Web community, cluster ranking, integrated cohesion, partitioning

摘要： 针对现有Web社区发现方法存在的不足及其聚合程度的测量问题，以社区节点、边、结构为对象，研究Web社区聚合强度的测量方法，分析社区最大化目标函数，以解决社区最优划分及主题优化问题，并提出CRIC社区发现算法。在现有信息搜索软件工具包的基础上构建其应用系统，实验结果验证该算法的有效性及适用性，能快速、高效地完成对网络社区的划分，具有一定的理论及应用价值。

关键词: Web社区, 聚类等级, 集成聚合度, 划分