一种结合小世界模型改良的NMF社区发现算法

A COMMUNITY DETECTION ALGORITHM BASED ON SMALL WORLD MODEL AND NMF

社区发现是当前复杂网络与数据挖掘的热点,非负矩阵分解是社区发现的常用手段。针对当前非负矩阵分解的社区发现算法,为提高算法的准确率与可解释性,提出多阶邻居节点的概念,在小世界模型的基础上构建了规模可控的多阶复合信息矩阵,用后处理的方法减少了算法中随机因素带来的不稳定性。对于真实网络与人工网络的实验证明,新背景下的算法较原算法在性能上有一定的提升。

Community detection is the hotspot of current complex networks and data mining,whose common means is non-negative matrix factorization. To improve the accuracy and interpretability of community detection algorithm,we propose the concept of first-order neighbors. On the basis of the small-world model,this paper constructed a controllable scale multi-stage compound information matrix. Treatment reduced the algorithm after using random factors of instability.Regarding experimental proof of the real network and artificial networks,new algorithms increase in performance compared to the original algorithm.