Study Of Community Structure Detection Algorithm In Complex Networks

Complex networks describe the relationships between interconnected units in systems such as society,biology,and information.Community detection algorithms,an important aspect of network analysis,can effectively identify characteristics of community structures,aiding researchers in understanding connections within these complex systems.In recent years,scholars from various research fields have proposed numerous algorithms for detecting community structures in complex networks,providing powerful tools for understanding their structure,function,and dynamic evolution process of complex networks.However,certain traditional community detection algorithms may have limitations such as node uncertainty,prior parameter requirements,and high computational costs.Additionally,further exploration is needed in combining community structure detection with node centrality and similarity.Therefore,this paper focuses on studying community detection algorithms using special structures in complex networks,cliques,and considering node centrality and local similarity.The specific work centers on two main aspects:(1)This paper proposes a community detection algorithm based on maximal clique of network nodes to study non-overlapping community structures in undirected and unweighted real networks.The algorithm first extracts the maximal cliques of nodes and then employs local similarity and clique grouping relationships between the proposed maximal cliques to perform cluster and merge hierarchically,effectively discovering the community structure within the network.To address issues arising from single-neighbor nodes and overlapping nodes during the merging process,we propose a modularity membership optimization criterion to determine their community membership relationships.We evaluate that the effectiveness of the algorithm is verified through three metrics on five real network datasets.(2)This paper proposes a community detection algorithm based on local gravity.The key strategy of the algorithm is to consider the network as a gravitational system and combine local similarity with a local gravity model to design a similarity mixing criterion,namely local gravity.Firstly,the algorithm utilizes a local gravity model to determine the central node in the network.Secondly,based on the local gravity criterion and the similarity between nodes,the algorithm detects the local communities within the network.Finally,the community structure is further explored through the local gravity effect.The experimental results demonstrate that the algorithm has good performance in detecting community structures on both real and artificial networks.