Most algorithms for local community detection select a seed node using a greedy algorithm and then expand it into a community using optimization functions. This paper presents a novel approach to community detection based on local expansion. The proposed Local Community Detection via LOcal Structure Expansion (CLOSE) algorithm features a novel connective function, which identifies a source node in the center of a highly-connected component of a graph. The CLOSE algorithm also selects a group of nodes rather than a single node as a seed for local community expansion, which facilitates the selection of a community suitable for the hub node. We also developed a system by which to identify the most suitable source nodes for given target nodes, referred to as Exploring Local Communities of Target Nodes (ELCTN). The performance of CLOSE and ELCTN was compared with that of state-of-the-art methods in experiments using synthetic networks generated using the Lancichinetti-Fortunato-Radicchi benchmark as well as real-world networks. Both algorithms outperformed previous methods in terms of accuracy and modularity.
All Science Journal Classification (ASJC) codes
- Human-Computer Interaction
- Hardware and Architecture
- Library and Information Sciences
- Computational Theory and Mathematics