Optimizing Subgraph Retrieval and Matching with an Efficient Indexing Scheme

Abstract

Graph index as an effective data structure is widely applied in subgraph retrieval and matching. It records and compares the frequencies of a set of specific features to detect subgraph containment on the fly, which is the foundation of the filtering techniques for subgraph retrieval and matching. However, due to the NP-hardness of the subgraph counting, current graph indices struggle to be built on large graphs. Even counting the simple path and cycle graphs is NP-hard. We observe that the monotone property of the counting process is crucial for the correctness and precision of the index. Therefore, we introduce an efficient graph indexing scheme by counting the path and cycle features monotonically in relaxed semantics. In addition to the filtering techniques, we propose to reorder the search candidates via our index. Experimental results reveal that our index can be constructed significantly faster than existing methods, by 1-3 orders of magnitude, and can handle graphs that are larger than previous work by 1-3 orders of magnitude. Our index-boosted filtering and ordering techniques are proven to be effective in optimizing the subgraph retrieval and matching process.