Evolving subgraph matching on temporal graphs

Abstract

Given a set of query graphs and specified orders between these query graphs, the evolving subgraph matching problem finds all subgraph sets from the temporal graph that not only match the given query graph set but also satisfy the specified orders. This problem motivates many significant application, e.g., social network analysis, financial fraud crimes detection, biological information mining, and so on. To efficiently address this problem, an evolving subgraph matching algorithm based on two key techniques is proposed. First, a memory-efficient index structure called EMI is designed to compactly store all necessary information required for finding all evolving matchings. To make the EMI compact and to construct the EMI efficiently, several techniques, such as four well-designed filters and adaptive filling orders, are proposed. Second, an evolving subgraph matching enumeration method is designed to immediately run on the EMI instead of on the temporal graph. The enumeration method expands evolving matchings in an edge-by-edge manner. Since the EMI is compactly maintained in the main memory and generally smaller than the temporal graph, the enumeration method runs very fast on the EMI. An extensive experimental evaluation has been carried out to evaluate the proposed algorithm. The experimental results show that the enumeration method executed on the EMI is an order of magnitude faster and requires significantly less memory than the baseline algorithms. To the best of our knowledge, there is no prior work on the evolving subgraph matching on temporal graphs.