种子节点贪婪扩张的重叠社区发现方法

Overlapping Community Discovery Method Based on Greedy Expansion of Seed Nodes

鉴于重叠社区发现通常具有复杂度高或结果不稳定的现象,提出了一种种子节点贪婪扩张的重叠社区发现方法.首先利用网络节点的拓扑特征寻找局部最大度节点作为种子,这些节点中心性好,且较好的分布在整个网络中,然后通过基于适应度函数的贪心策略扩张种子,并在每次有新节点加入社区时清洗社区,以此发现质量高的重叠社区.本文选取了人工模拟网络和真实网络进行了对比实验,实验结果表明,该算法能发现较高质量的重叠社区结构.

In view of overlapping community discovery,it is usually complicated or unstable,an overlapping community discovery method based on greedy expansion of seed nodes was proposed.First,the paper employed the topology features of network nodes to find local maximum nodes which were used as seeds.These nodes have good centrality and better distribution in the whole network.Then,the seeds were expanded by the greedy strategy based on fitness function,and a high quality natural community was found at each time when a new node joined the community.This paper chooses the artificial simulation network and the real network to carry on the contrast experiment,the experimental result shows that this algorithm can discover the high quality overlapping community structure.