Research On Multiple Query Methods On Large Scale RDF Graph

With the development of artificial intelligence and the increasing demand for knowledge,knowledge mapping emerges as the times require.Most knowledge maps use RDF to describe resources.SPARQL BGP query on RDF graph is an important technology to describe,mine and analyze all kinds of knowledge maps.SPARQL BGP query on RDF graph is equivalent to subgraph matching problem.Since subgraph matching problem has been proved to be NP problem,how to efficiently query SPARQL BGP on large-scale RDF graph is a challenging problem.In this paper,we design multiple query methods on large-scale RDF graph for semantic queries with different query constraints.Firstly,when executing SPARQL query constrained by predicate constant,we use the distributed storage of star like structure and tag tree index based on predicate constant in this paper to perform query graph preprocessing and star like structure matching process.By reducing the number of query iterations and index efficient pruning filtering effect,we can speed up the retrieval speed,and use E-MJOC algorithm based on star like structure The matching and join order of query subgraphs are determined to optimize the query,and the join cost is reduced by reducing the generation of intermediate results,so as to improve the query efficiency.Secondly,when SPARQL query with mixed constant constraints is executed,the node partition index and coding tree index based on Bloom filter are designed.When preprocessing query graph and matching star like structure,the calculation range is reduced according to the cut point information and node partition index in RDF data to avoid unnecessary calculation of irrelevant storage nodes,and then the coding based on Bloom filter is used The tree index can prune and filter effectively,reduce the number of candidates to be matched,and reduce the cost of matching.Finally,the MS-MJOC algorithm is used to generate the corresponding matching and connection query plan,and the intermediate results are connected based on the common points,which improves the overall query efficiency by reducing a large number of connection operations.Finally,different RDF datasets are used for experimental comparison.The experimental results show that the SPC storage and indexing method and the optimization method of the query itself can promote the semantic query constrained by predicate constant.On the other hand,our SMC performs better than SDec and S2 X in semantic query with mixed constant constraints.