基于中心差异度迭代调整机制的 网络社区搜寻算法研究

为解决当前网络社区搜寻算法存在的节点聚类形成困难,搜寻迭代过于复杂,难以实现社区归属的二次更新等不足,提出了一种基于中心差异度迭代调整机制的网络社区搜寻算法。首先,通过领袖节点重叠度来实现初次社区搜寻裁决,有效降低了重复搜寻的概率,且根据加入节点与领袖节点差异度进行聚类匹配;随后,通过待加入节点与领袖节点之间的交互热度方式进行基于热度机制的聚类递归,实现对搜寻误差的二次校正。仿真实验表明,与当前网络社区搜寻算法中常用的差分迭代阈值裁决机制,混沌度一体化成型迭代机制相比,本文算法具有更高的首次成功率,以及更小的搜寻次数与迭代周期,具有很强的实际部署价值。

THE RESEARCH AND SIMULATION OF NETWORK COMMUNITY SEARCH ALGORITHM BASED ON CENTRAL DIFFERENCE ITERATION ADJUSTMENT MECHANISM

In order to solve the problems of node clustering in current network community search algorithms,such as the complexity of search iterations and the difficulty of secondary update of community ownership,a network community search algorithm based on the iterative adjustment mechanism of center difference degree is proposed.Firstly,the overlapping degree of leader nodes is used to decide the initial community search,and the difference between join nodes and leader nodes is matched to improve the speed of clustering formation and reduce the search iteration process.In order to improve the accuracy of community discovery,the method of constructing the interaction heat between the node to be added and the leader node is used to improve the accuracy of community discovery.Simulation results show that the proposed algorithm has higher first success rate in comparison with the Differential Iterative Threshold Decision Mechanism and Chaos Integration Forming Iterative Mechanism which are commonly used in current network community search algorithms.The advantages of fewer searches,shorter iteration period and clearer aggregation degree make it valuable for practical deployment.