An Efficient Algorithm For Detecting Overlapping Community On Complex Network

The overlapping community detecting algorithm of complex networks is used to solve the problem of identifying overlapping communities in complex networks,which helps to acquire and understand the overall structural characteristics of complex networks.Since one node in the overlapping community detecting may belong to multiple communities,the nodes belonging to multiple communities cannot be identified by the traditional community detecting algorithm.With the development of society and technology,the scale of complex networks is getting larger and larger,and the existing overlapping community detecting algorithms are less efficient in dealing with such large-scale networks.Therefore,faster algorithms are needed to better cope with overlapping community detecting problems in large complex networks.This paper proposes an algorithm for selecting the seed in the whole network and then locally optimizing the seed through the income function to discover the overlapping community structure of the complex network.The work is mainly divided into two parts.First,the seeds are selected in the whole network.This paper proposes two seeding strategies: “community expansion” and “high-degree node expansion”.The “community expansion” seeding strategy uses the Louvain algorithm to initially divide the network,and the resulting segmentation result is used as a seed set;the “high-degree node expansion” seeding strategy uses a priority-based approach to select seeds.It should be noted that the degree used in this paper takes precedence.The way the seed is selected takes into account the effect of the selected seed on the subsequent seed selection,and the degree of the remaining nodes is updated each time a seed is determined.Then,the seed in the seed set is sequentially optimized by the income function,and expanded from each seed.The income function is used to determine whether the neighborhood node can join the community and whether the internal nodes of the community are deleted until the income function reaches local optimum,and the community structure is determined.The algorithm proposed in this paper avoids searching the entire network in each iteration of the detecting community.This paper theoretically analyzes the time complexity of the algorithm and experiments on the real data set.It verifies the performance of the algorithm from the running time of the algorithm and the accuracy of the experimental results,and compares it with Big Clam algorithm,OSLOM algorithm and DEMON algorithm.The experimental results show that the algorithm accuracy is slightly lower than that of the compared DEMON algorithm when the seed is selected by the Louvain algorithm,but the algorithm is more efficient than the three algorithms compared,and is significantly higher than the Big Clam algorithm..When the seed is selected by the usage priority policy,the running efficiency and accuracy of the algorithm are higher than the three algorithms compared.The algorithm proposed in this paper can accurately detect overlapping communities in complex networks,and has the advantages of high efficiency.It can be used for large-scale complex network overlapping community detecting.