Size-Fixed Group Discovery Via Multi-Constrained Graph Pattern Matching

Guliu Liu,Lei Li,Guanfeng Liu,Xindong Wu
DOI: https://doi.org/10.2139/ssrn.4361657
IF: 8.1
2024-01-01
Information Sciences
Abstract:Multi-Constrained Graph Pattern Matching (MC-GPM) seeks to match a pattern graph with multiple attribute constraints on nodes and edges, and has received increasing attention in different applications such as social-based e-commerce, trust-based group discovery, etc. However, the existing MC-GPM methods do not consider situations where the number of each node in the pattern graph needs to be fixed, such as finding experts group with quantities and relations identified. In this paper, a Multi-Constrained Strong Simulation with the Fixed Number of Nodes (MCSS-FNN) matching model is proposed, and then a Trust-oriented Optimal Multi-constrained Path (TOMP) matching algorithm is designed for solving it. Moreover, a heuristic optimization strategy for combinatorial testing and an optimization strategy for edge matching are designed to improve the efficiency of TOMP algorithm. Finally, empirical experiments are carried out on four real social network datasets, and the experimental results demonstrate the effectiveness and efficiency of our proposed algorithm and optimization strategies.
What problem does this paper attempt to address?