基于子团规模的社团划分算法与地理位置

针对以往社团划分算法中存在的子团规模过大的问题,在CNM算法的基础上重新定义子团规模,提出了CLCNM社团划分算法.社团划分结果表明,改进后的算法在子团数量和模块度方面要优于CNM算法.在此基础上,研究社团划分与地理位置间的关系,提出地址前缀相似度的概念,子团地址前缀相似度越大,说明社团划分后的地理效应越明显.CLCNM社团划分结果表明:IPv6网络子团具有明显的地理效应,即子团中节点分布在相邻的地理位置.这一结论可对网络拓扑的再部署提供借鉴性意见.

Partition Algorithm Based on the Size of Sub-community and Geography

A new definition for the size of community was proposed based on the CNM algorithm, which was to solve the problem of excessive size of community exist in the past algorithm for detecting community. The new partition algorithm was called CLCNM. The results of community partition showed that the improved algorithm was superior to the CNM algorithm on the number of community and modules. On this basis, the relation between community partition and geography was studied, and a conception of the similarity of address prefix was proposed, which indicated that the greater similarity, the more obvious effect of relation between community and geography. The results of community partition by CLCNM showed that the communities of IPv6 topology had an obvious effect of geography, that was the sub-group node located in adjacent geographical location. This conclusion can provide advice to the re-deployment of network topology.