主要从节点在所属群体内的相对重要程度出发,尝试性地给出一种网络社团结构的新定义,并在此基础上构建了一种网络社团划分的评价及改进方法。该方法既可以与现有的社团划分算法进行组合,形成新的网络社团划分算法,也可以独立使用,对网络社团的划分结果进行评价与改进。最后,通过MCL、GN、Factions等方法及算例对提出的算法进行了验证与分析。
This paper proposed a new definition of community structure from the perspective of the nodes' relative impor- tance. Based on the theory of degree centrality and standardized degree centrality, it mainly researched the evaluation and im- provement method on division of community structure in network. The new method could be combined with other community di- vision algorithms to form new community division algorithms. The new method Could also be used independently to evaluate and improve the community division results in network. At last, it illustrated the new evaluation and improvement method through MCL algorithm, GN algorithm, Factions algorithm and examples.