Large-scale complex network oriented dense overlapping community division method

Abstract

The invention relates to a large-scale complex network oriented dense overlapping community division method. The method comprises the following steps: S1, abstracting a large network into an undirected graph, initializing system parameters, setting initial community circle set as a null set, and setting iteration stop conditions; S2, traversing nodes that are not accessed, setting the nodes as seeds of community circles, and expanding the circle from the nodes; maintaining an extensible node set of the nodes by using a priority queue, maintaining neighbor nodes by the priority queue, and continuously updating; S3, continuously extending the nodes, completing community extension when conductance of the community circles cannot be reduced any more or reaches the iteration stop conditions, and adding the community circles to the community circle set; S4, judging whether all the nodes are completely accessed or not, and if not, jumping to the S2, otherwise entering S5; S5, outputting the community circle set, and ending the algorithm.

Description

technical field [0001] The invention relates to the fields of parallel computing and social networks, and more specifically, relates to a method for dividing dense and overlapping communities for large-scale complex networks. Background technique [0002] In recent years, with the rise of large social networks (such as Facebook, Twitter and Weibo), community detection (Community Detection) has gradually attracted widespread attention from academia and industry. The goal of community mining is to dig out social circles with high cohesion and low coupling from the abstract network. These circles may be dense and overlapping. At present, Facebook has more than 2 billion monthly active users, so many users have built a huge social network. Community mining can bring a lot of benefits. For example, Facebook can recommend friends to members in circles with the same hobbies; Amazon can mine circles with similar purchasing interests based on its network of goods and users, and reco...

AI Technical Summary

Problems solved by technology

[0003] However, there are still many challenges and deficiencies in the current community mining research.

The first is to solve the computational complexity problem of large-scale networks. Existing researches believe that mining high-cohesion and low-coupling community circles in huge social networks requires the design of algorithms with low complexity. Therefore, it is necessary to design Good heuristic algorithm; second, in the mining research of dense overlapping communities, existing research often assumes that the number of mined communities is already known, but this number is generally difficult to estimate and can only be predicted based on experience. Third, in the existing research, researchers often cut the large-scale network abstract graph into several small graphs, which will lose the information of the original graph, although some studies have explored how to Minimize the loss of information in the graph and cut it into several separate graphs, but the useful information of the original graph will always be lost

Images