Querying RDF graphs over partitioned indexes
Lei Gai, Junmin Liu, Xiaoming Wang, Jian Li · 2017
We consider the problem of improving the query response time over a large-scale RDF collection. Existing methods need to maintain either all intermediate results or all graph structured data, which result in efficiency problems for complex queries. Our main idea is to introduce a two-level processing. The first level is performed as subgraph matching over a partition-based summary graph, which output coarse-grained matched partitions in a stream fashion. Then the fine-grained joins are processed on the matched partitions and generates the final results. Our method is efficient in that it avoid processing on large SPO permutation index at query time. An extensive experimental study using benchmark and real dataset shows that the query performance of our approach outperforms existing systems by orders of magnitudes.