Study On Network Overlapping Community Detection By Signaling Process Based On True Line Graph Framework

Network community structure is one of salient features in complex networks,community detection can help explore the functional structure and dynamic characteristic of the network,based on which we can conduct typical application research.Most existing community detection algorithms tend to divide the network to disconnected completely communities,yet communities in real-world networks maybe overlapped with multiple properties of the nodes.Overlapping communities not only reflect truly the topology structure of the network,but also play an important role in interaction between communities.Link community partition is a contrary thinking in community detection and fly its own colors in community detection research with its novel perspective,but its drawbacks are low efficiency and poor accuracy.In order to make the best of the existing non-overlapping community detection algorithms,we proposed a network overlapping community detection algorithm framework driven by true link.The concept of “true-link” graph is firstly proposed in this paper and used to pre-processing the original network;link communities are obtained through dividing the line graph of “true-link” graph with a non-overlapping community detection algorithm;community structure is produced by reduce the link communities to the node communities.A LPA network overlapping community detection algorithm driven by true link is proposed based on the algorithm framework and the testing result on karate network shows that the proposed algorithm can detect community structure of the network effectively.Community detection is a clustering problem.A network overlapping community detection by signaling process based on true line graph framework is proposed combined with signaling process and true line graph framework.Signaling process and single-linkage hierarchical clustering are applied to the true line graph framework;in the process of reducing link communities to node communities,both communities will be merged if they are very similar;the iterative calculation is continuing until the modularity is maximized.Seven real networks and two groups of artificial networks are employed to test the performance of TLCCD,and the testing results show that the algorithm can overcome the defect of link partition that the redundant overlapping communities are partitioned to some extent,meanwhile obtain usually the community structure of higher accuracy.The research work of this paper is still exploratory in the field of community detection.The proposed algorithm framework driven by true link and signaling process based on true line graph framework are active and useful to the research of overlapping community detection.