Semantics Preserving MapReduce Process for RDB to RDF Transformation

Abstract

Today, most of the data on the web is stored in relational databases, which is called deep web. Semantic web is a movement to the next generation of the web, where all data are augmented with well-defined semantics and linked together in machine-readable format. RDB2RDF approaches have been proposed and standardized by W3C, which publishes relational data to semantic web by converting relational data into RDF formatted data. We propose a system that automatically transforms relational data into RDF data and creates OWL ontology based on the schema of database. Some approaches have been proposed, but most of them did not fully make use of schema information to extract rich semantics, nor did they experimented on large databases for performance. We utilize Hadoop framework in transformation process, which enables distributed system for scalability. We present mapping rules that implements augmented direct mapping to create local ontology with rich semantics. The results show that our system successfully transforms relational data into RDF data with OWL ontology, with satisfactory performance on large-sized databases.