Communications in Information and Systems

Volume 22 (2022)

Number 1

G-Explorer: a visual exploration and analysis method of subgraph in complex network based on graph embedding

Pages: 147 – 167

DOI: https://dx.doi.org/10.4310/CIS.2022.v22.n1.a7

Authors

Qinghui Zhang (Beijing Key Laboratory of Big Data Technology for Food Safety, School of Computer Science and Engineering, Beijing Technology and Business University, Beijing, China)

Yi Chen (Beijing Key Laboratory of Big Data Technology for Food Safety, School of Computer Science and Engineering, Beijing Technology and Business University, Beijing, China)

Menglu Zhang (Beijing Key Laboratory of Big Data Technology for Food Safety, School of Computer Science and Engineering, Beijing Technology and Business University, Beijing, China)

Zeli Guan (Beijing Key Laboratory of Big Data Technology for Food Safety, School of Computer Science and Engineering, Beijing Technology and Business University, Beijing, China)

Abstract

As the scale and complexity of networks increase, the mining and analysis of subgraphs in complex network faces great challenges. Graph embedding technique can transform the high-dimensional massive subgraph data into low-dimensional computable embedding vectors while preserving the structural features of subgraphs. In this paper, a graph embedding-based visual exploration and analysis method, G-Explorer, is proposed to help users analyze selected subgraphs from multiple perspectives. The method first constructs a vector representation of the nodes using struc2vec, and then clusters the vectorized nodes to form subgraphs using KMeans. A new visualization design, Force-Radar, is also proposed, in which radar plots reflecting the characteristics of subgraphs are used as nodes and the associations between subgraphs are used as edges to form an overview of the whole network. Detailed information about the selected subgraph is presented in multiple auxiliary views through interaction. The method was applied to analyze the association network of food inspection data for exploring the high-risk foods and hazards. The experimental results illustrate the effectiveness of the G-Explorer method.

The research of Y. Chen was supported is part by National Natural Science Foundation of China (61972010), National key R&D program of China (2018YFC1603602) and Basic Research Project of the Ministry of Science and Technology (2015FY111200).

Received 25 October 2020

Published 7 February 2022