Label propagation overlapping community discovery method based on node comprehensive influence

Abstract

The invention relates to the technical field of big data mining, and discloses a label propagation overlapping community discovery method based on node comprehensive influence, which comprises the following steps: calculating node importance LI in a network according to topological information: degree, triangle number and local mean of nodes; calculating a node similarity influence SI according to the similarity Sim of the nodes and neighbors thereof and the node intimacy NI; calculating the comprehensive influence CI of the nodes according to the node importance LI and the similarity influence SI, sorting all the nodes according to the node importance LI, initializing a unique label and a main label for each node, and reserving and normalizing the labels meeting a threshold value; and checking all the labels, deleting the labels which do not meet a node number threshold value, retaining the nodes in the labels, calculating the membership coefficient again, and redistributing the community. According to the method, the labels of the nodes are updated asynchronously, the ending condition is that the main labels are not changed twice or reach the maximum iteration value, and the overlapped nodes and community structures in the complex network can be efficiently and accurately found.

Description

technical field [0001] The invention relates to the technical field of big data mining, in particular to a method for discovering overlapping communities of tag propagation based on the comprehensive influence of nodes. Background technique [0002] There are many complex network data in the real world, such as infectious disease transmission network and social network data. These complex networks are characterized by "small-world" and "scale-free", and structural features include degree, shortest path length, and betweenness of nodes. In addition, there are deeper structural information in complex network data. This community structure implies that a network is divided into several groups such that nodes within each group are densely connected, while nodes between different groups are sparsely connected. Nodes belonging to the same community have similar characteristics or close connections. At present, many scientists realize that mining community structure from these h...

AI Technical Summary

Problems solved by technology

It is very difficult for these algorithms to detect communities in networks with more than 100,000 edges

At present, there are also overlapping community discovery algorithms based on label propagation by some researchers, but most of the previous label propagation discovery methods can only be used to detect non-overlapping communities.

Images