Multi-mode heterogeneous social network community discovery method

Abstract

The invention provides a multi-mode heterogeneous social network community discovery method. The method comprises the following steps: constructing an initial information matrix based on information of each node in a multi-mode heterogeneous social network and a meta-path connection relationship set among different nodes; dividing the initial information matrix by taking the shortest reachable meta-path as a division basis to obtain a multi-dimensional heterogeneous network information matrix; performing dimension combination on the small information vectors and the overall information vectors of at least two preset target nodes in the multimode heterogeneous social network to obtain an information matrix of a preset target node set; calculating a fusion vector of the initial information matrix; and acquiring a community set according to the fusion vector. According to the method, a meta-path data storage mode is adopted, so that a relation chain is formed by relations between users and attributes, more excellent knowledge expression ability is further obtained, principal components in the path are extracted by utilizing a factor analysis model, the current situation that node attribute quantification is incomplete is improved, and the cohesiveness of a community structure is improved.

Description

technical field [0001] The present application relates to the technical field of community discovery, and in particular to a multi-mode heterogeneous social network community discovery method. Background technique [0002] Community detection is an important research problem in social network analysis. Generally speaking, the community is a dense subgraph in the network, which satisfies the characteristic that the links between the nodes in the community are relatively close, and the links between the nodes in the community are relatively sparse. Due to this characteristic, the community structure can well describe the clustering characteristics of the network, which has a positive effect on application systems such as epidemic prevention and control, community management and control, and personalized recommendation, and greatly improves the performance of the system. In traditional social networks, methods only consider the similarity between nodes to identify communities....

AI Technical Summary

Problems solved by technology

[0003] In summary, the main problems of existing community discovery methods are: only for homogeneous information networks, ignoring the expansion of optimization objectives caused by the heterogeneity of nodes and links; storing graph data in matrices or adjacency lists, nodes are Represented as an adjacency list entry or a matrix element, the limitations of this data structure lead to the inability to effectively store heterogeneous social networks with node attributes; when dividing communities, only "node pairs" are used as the basis for division, Unable to discover the hidden joint attributes behind the links, resulting in poor cohesion of the divided communities and lack of practical significance

Images