Multi-view graph clustering method based on graph data

Abstract

The invention discloses a multi-view graph clustering method based on graph data. The method comprises the following steps: firstly carrying out the filtering of each view feature of the multi-view graph data through a low-pass filter, and then calculating a similar matrix through a self-expression model with good performance in combination with a weight mechanism; according to the method, a similar matrix and an adjacency matrix representing a graph node relation are associated by setting a regular term to obtain an iterative algorithm, a graph adjacency matrix of optimal representation of multi-view graph data is obtained through quick convergence of the algorithm, and a final clustering result can be obtained by applying a traditional spectral clustering method to the matrix. The method is realized based on a graph filtering technology, high-order adjacent information of a graph structure matrix and a weight mechanism, compared with a method based on deep learning, the method is simple and efficient, avoids a calculation process of a large number of parameters, and has good performance indexes on the premise of ensuring efficiency; and moreover, the universality is high, and compared with the existing method, the method has huge advantages on a plurality of widely applied data sets.

Description

technical field [0001] The invention belongs to the field of graph data processing, and in particular relates to a graph data-based multi-view graph clustering method. Background technique [0002] With the rapid development of the Internet and social media, network data has become ubiquitous. They are naturally represented as graphs, a typical non-Euclidean data structure. Graph clustering, also known as community detection or graph partitioning, aims to group nodes in a graph into distinct clusters. Due to its wide range of applications, graph clustering has been extensively studied. Most existing graph clustering methods mainly focus on the graph topology. In practical applications, real data usually contains a description of each node and pairwise structural relationships between nodes. To exploit rich node information, some recent techniques combine attributes and graphs. The results show that integrating attributes and graphs can help improve clustering performanc...

AI Technical Summary

Problems solved by technology

However, although they are effective for processing single-view data, they are not suitable for processing multi-view graphical data

Images