SPARQL-to-HiveQL Translation

Abstract

Resource Description Framework (RDF) is a W3C Recommendation for knowledge representation on semantic web. Growing size of RDF anno- tated data demands scalable semantic stores. Hadoop based distributed and parallel processing frameworks such as HBase and Hive are increasingly be- coming popular for storing voluminous data and for enhancing flexibility to handle complex data. Hive is a Hadoop based data warehousing infras- tructure with support for complex analytical processing. Its query interface doesn’t support data exploration using SPARQL, a standard query language for RDF. Integration of aforementioned technologies with added support for SPARQL queries may realize a scalable semantic web data store. We have proposed a semantic preserving SPARQL-to-HiveQL translation scheme that adds querying interface to the Hadoop based RDF stores. Major contribu- tions of our research are (i) semantic preserving SPARQL-to-HiveQL query translation algorithm (ii) a storage schema independent querying mechanism that accommodates different storage schemes without impacting translation time. The experimental results show the efficient working of proposed trans- lation algorithm and its support for different types of SPARQL queries.