Furthermore, this article proposes a principle of designing community detection algorithm (the principle which states the relationship among algorithms, benchmarks and the reality of community partition) and offers the details about designing model. Depending on the recommended designing model, this article proposes a relatively new method to demonstrate how to devise and improve detection algorithms flexibly using the proposed model. Some test results showing the effect of CNM and proposed algorithm are also provided in our tests, including a trial on a real small-scale network and tests on artificial networks.  
Keywords    community detection, community partition, clustering algorithm, agglomerating method, algorithm analysis  复杂网络社团结构发现算法研究与实现(2):http://www.751com.cn/jisuanji/lunwen_65828.html