Community recognition method in social network

Abstract

The invention provides a community recognition method in a social network. The method comprises the steps of analyzing a time evolutionary structure of a social network graph, encoding a network evolutionary model formed by various possible solutions in a problem solving process, converting a social network into a uniform network structure with a low entropy value, calculating encoding cost of a social relationship, and searching an optimal solution through a clustering algorithm to obtain a stable community structure. The invention provides a community recognition method in a social network.When the method is applied to a network with link loss or noise data, the influence of network mutation and accumulation errors on community discovery precision is effectively reduced, the influence of node direct neighbors and the influence of indirect neighbors are considered, the sensitivity of an algorithm to neighborhood threshold parameters is reduced, and the accuracy of a community attribution result is effectively improved.

Description

technical field [0001] The invention relates to big data, in particular to a community identification method in a social network. Background technique [0002] Users in social networks form communities due to common interests and social attributes. Communities are usually composed of network nodes with similar functions or properties, which to some extent reflect the local weak regularity and global order behind the spontaneous behavior of users. The discovery and recognition of the community structure of social networks helps to reveal how interrelated users form complex social networks, and has many practical values, such as obtaining groups with common preferences or similar social backgrounds by discovering communities in the network. At present, most community discovery technologies are oriented to automatic community discovery, that is, according to the network structure, automatically discovering communities in the network through algorithms is an unsupervised method...

AI Technical Summary

Problems solved by technology

However, there are usually two important problems in automatic algorithms: the network obtained by users contains a lot of noise data including wrong connections, and the network connection is usually incomplete. The performance of automatic community discovery will be greatly reduced at this time, and it is difficult to effectively Identify the real community structure in the network; and discover the community by maximizing the modularity through the approximate optimization method. Even if the modularity value is very small, it may lead to very different community divisions, and an optimal division result cannot be obtained. Difficulty finding community structures that meet specific needs in specific contexts

Images