Statistics and Its Interface
Volume 12 (2019)
Spectral clustering-based network community detection with node attributes
Pages: 123 – 133
Identifying communities is an important problem in network analysis. Various approaches have been proposed in the literature, but most of them either rely on the topological structure of the network or the node attributes, with few integrating both aspects. Here we propose a community detection approach based on spectral clustering combining information on both the network structure and node attributes (SpcSA). Some of the attributes may not describe the communities we are trying to detect correctly. These irrelevant attributes can add noise and lower the overall accuracy of community detection. To determine how much each attribute contributes to community detection, our method introduces a mechanism by which attribute weights can adjust themselves. We demonstrate the effectiveness of the proposed method through numerical simulation and with real-world data.
spectral clustering, community detection, stochastic block model, node attributes, normalized mutual information
2010 Mathematics Subject Classification
Primary 62-07. Secondary 68U20, 91D30.
The project was sponsored by the National Natural Science Foundation of China (11301236), the Natural Science Foundation of the AnHui Higher Education Institutions of China (KJ2017A377, KJ2017A376), and the Anhui Provincial Natural Science Foundation (1608085QG169).
Received 13 November 2017
Published 26 October 2018