Abstract
Subgraph matching is an important research topic in the area of graph theory and it has been applied in many areas in nowdays. Filtering and verification are two main processes of subgraph matching algorithms. However, there exists many invalid nodes in candidate matching set after initializing the candidate set for each query node, which may result in a quantity of redundant computation during the filtering period. Regarding the problem mentioned above, in this paper, we propose a subgraph matching algorithm based on node influence, denoted as InfMatch, to improve the performance of subgraph matching on a large target graph. Specially, we find the central node of query graph by calculating the global and local influence value of each query node, after which candidate matching nodes for each query node are found from the neighborhood region of the candidate nodes for the central node. Since the central node we choose connects tightly with other nodes, isolated nodes can′t be added into the candidate matching set for central node and thus a number of unqualified candidate vertices are pruned. To further prune the unqualified candidate nodes, we propose several filter strategies according to the characteristics of our method. What′s more, considering edge limitation, we improve the matching order selection strategy. Extensive experiments demonstrate that our method is more efficient.
•Propose a method based on the influence value of the vertices to determine the matching order.•Propose a extending method based on the central node for generating a sub-areas.•Propose a subgraph matching algorithm to find the candidate nodes on sub-areas instead of the whole target graph.