A Twig-Based Algorithm for Top-k Subgraph Matching in Large-Scale Graph Data

Abstract

Subgraph matching is considered as a basis query for graph data management, and is used in many domains, such as semantic web and social network analysis. Subgraph isomorphism is an initial solution for the task, which is an NP-complete problem. To speed up the procedure, graph simulation has been presented to match subgraphs with polynomial complexity. Unfortunately, simulation usually loses topology of matched subgraphs. In this paper, we propose an approximation approach for subgraph matching based on twig patterns. First, we transform query graphs into twig patterns and match candidate substructures in graph data. Second, we present an optimized join strategy along with top-k mechanism, including join order selection based on cost evaluation and optimized pruning based on maximum possible score and minimum possible score. Finally, we design experiments on real-life and synthetic graph data to evaluate the performance of our work. The results show that our approach obviously reduces the time complexity and guarantee the correctness for answering the queries of subgraph matching.