[摘要] 提出了一种基于权重属性的图聚类方式。该图聚类方式在图聚类的基础上,考虑了每个节点的不同属性,并根据影响度给属性分配权重,从而在依据亲密度构建的网络拓扑图上进行图聚类的修正。实验证明,该方法更符合实际的群体聚合方式。
[关键词] 社群挖掘;图聚类;相似度计算
[Abstract] This paper proposes a graph-clustering algorithm based on attribute information. The attributes (and their weights) of each node are considered in this model when modifying the network topology based on intimacy. Experiments show that the modified algorithm is closer to the actual group polymerization.
[Keywords] community detection; graph clustering; similarity calculation