Efficient $K$-concern Matching in a Large Graph

Exact subgraph matching is wildly used for relation exploration in linked data. Existing literatures of matching algorithm focus on exhaustive enumeration of all matches, which are both time and space consuming and even not viable for a large graph. For real world applications such as social search, it is common that not all entities in a query graph are concerned by the users. It is not trivial to find distinct groups of concerned vertices from the exhaustive enumeration of all matches. In this paper, we define the problem of $K-concern \ \ matching$ which returns distinct groups of $K$ concerned vertices in all matches. In order to efficiently find $K$ -concern matches in a large graph, we devise a new strategy for the general backtracking algorithm. To minimize the total search space, we propose methods for search order optimization and candidate set pruning. Experiments on real and synthetic datasets show that the query performance of our method exceeds all existing matching algorithms for $K-concern \ \ matching$.