Query Rewriting from SPARQL to SQL for Relational Database Integration

Integrating heterogeneous relational databases into a Semantic Web is the essential step to support the precise semantic search and query, where query rewriting from SPARQL to SQL is the key technique. This paper presents an approach to rewriting a SPARQL query to SQL queries based on semantic mappings from relational schemas to a universal shared ontology. The ontology schema is used as the mediation layer to eliminate the heterogeneities of large scales of databases by mapping relational schemas to it. A visualized mapping editor is well developed to help users define mappings. The query rewriting approach is divided into three steps: (1) define mappings from relation schemas to the universal ontology by using the editor; (2) preprocess the mappings and the SPARL query for the rewriting; (3) rewrite the SPARQL query to SQL queries based on the preprocess results. In addition, this approach supports rewriting SPARQL queries containing operators ?optional‘, ?filter‘ and ?union‘. Its effectiveness is demonstrated using a real use case. In particular, this approach has been well developed and in-use in a traditional Chinese medicine (TCM) semantic search and query platform.