ohStore: Ontology hierarchy solution to improve RDF data management

RDF data is growing exponentially and expanding rapidly from different sources everyday. Storing large-scale RDF data has been a matter of concern since the beginning of the Semantic Web. In the recent past, multiple approaches for storing RDF data have been suggested, ranging from simple storages to advanced methods like clustering predicates, which belong to a single class, or vertical partitioning on the properties. Unfortunately, it is still a challenge to store and retrieve huge quantities of RDF triples, due, in part, to the unpredictable nature of data encoded in RDF. Current RDF stores in RDB scale poorly, which may exacerbate poor performance behavior for querying and retrieving RDF triples. In light of this, we propose a framework, called ohStore, which improves state-of-the-art scaling methods by developing a storage model based on the ontology hierarchy and by clustering classes with the help of their semantic relatedness. Our experiment shows that ohStore achieves encouraging and competitive performance by reducing main bottlenecks of property-table, one of the state-of-the-art RDF storage method into RDBMS, on both row and column store database systems.