Mobile communication user group construction method based on faction filtering and label propagation

Abstract

A mobile communication user group construction method based on faction filtering and label propagation belongs to the field of data business. Calculate the closeness of connections between user nodes; build a weighted complex network between user nodes; based on the weighted complex network, use the faction filtering algorithm to construct the seed group of the social relationship group; based on the seed group, initialize all node labels in the network , that is, assigning an initial label to each node; using the improved SLPA algorithm for label propagation, when the labels of most nodes converge, the nodes are divided into corresponding groups based on the node labels to complete the construction of social relationship groups. That is, nodes with the same label form a group; the advantages of this invention are: it can obtain better user friend recommendation effects; it can obtain better collaborative recommendation effects; it is helpful for user composition analysis; it is helpful for discovering abnormal groups; next The division basis of a generation of communications.

Description

technical field [0001] The invention belongs to the field of data services, in particular to a mobile communication user group construction method based on faction filtering and label propagation. Background technique [0002] In the prior art, there is no document about the method of constructing a social relationship group through communication information between users and discovering a group center user. Relatively similar SLPA (Speaker-listener Label Propagation Algorithm, speaker-lisener label propagation strategy algorithm) algorithm has near-linear time complexity and good performance, which is more suitable for users with large data sets. The construction of social relationship groups among them. The SLPA algorithm initially assigns a unique label to all user nodes, which slows down the iterative convergence time of the algorithm, and may cause oscillations in the synchronous update strategy, and its robustness is relatively poor. In addition, although the algorit...

AI Technical Summary

Problems solved by technology

The SLPA algorithm initially assigns a unique label to all user nodes, which slows down the iterative convergence time of the algorithm, and may cause oscillations in the synchronous update strategy, and the robustness is relatively poor

In addition, although the algorithm can discover overlapping communities, it is mainly aimed at unprivileged networks and cannot be directly applied to authorized complex networks.

Images