ISSN 1000-1239 CN 11-1777/TP

Journal of Computer Research and Development ›› 2020, Vol. 57 ›› Issue (6): 1302-1311.doi: 10.7544/issn1000-1239.2020.20190572

Previous Articles     Next Articles

RGNE:A Network Embedding Method for Overlapping Community Detection Based on Rough Granulation

Zhao Xia1, Zhang Zehua1, Zhang Chenwei2, Li Xian1   

  1. 1(College of Information and Computer, Taiyuan University of Technology, Taiyuan 030024);2(School of Computer Science, University of Illinois at Chicago, Chicago, USA 60607)
  • Online:2020-06-01
  • Supported by: 
    This work was supported by the National Natural Science Foundation of China (61503273, 61702356) and the China Scholarship Council Program (201806935047).

Abstract: Community mining of complex information networks is a research hotspot in recent years and the detection of overlapping communities has important practical significance. The traditional community detection method accurately divides all nodes into each subclass to form a non-overlapping partition. However, the hard partitioning method is more difficult to deal with complex situations involving uncertain information and noise information. At present, there are few researches on the method of network embedding for overlapping community detection. Aiming at the problems of community drift and boundary uncertainty, a network embedding community detection method based on rough granulation is proposed. The node representation of structure information and attribute information is obtained through network embedding, and the similar nodes are mapped to the low-dimensional continuous vector space with similar distances. Then, the network structure and multi-level information with rough granulation to deal with the uncertainty areas are considered, and overlapping communities are finally generated. The experimental results in network public datasets and synthetic datasets show that the RGNE(network embedding based on rough granulation)method has higher precision and can effectively deal with community detection problems of uncertain networks. Finally, the parameter settings affecting the experimental results are discussed and analyzed in detail.

Key words: community detection, overlapping community, community drift, network embedding, rough granulation

CLC Number: