Title: An incremental and distributed inference method for large-scale ontologies over SPARK

Authors: Mohamed Oubezza; Ali El Hore; Jamal El Kafi

Addresses: Department of Computer Sciences, Faculty of Sciences, Avenue Ben Maachou, 24000, EL Jadida, Morocco ' Department of Computer Sciences, Faculty of Sciences, Avenue Ben Maachou, 24000, EL Jadida, Morocco ' Department of Computer Sciences, Faculty of Sciences, Avenue Ben Maachou, 24000, EL Jadida, Morocco

Abstract: The study of the semantic interoperability and the reasoning over big data is today a major challenge for researchers, especially with the birth of semantic web and deep data. The existing solutions are not yet able to satisfy the requirements of the final user especially in terms of the consistency of the results and the request execution time. To do this we need an approach based on an ontology and a distributed and scalable system. Several studies have been done on the reasoning over large-scale ontologies, most are based on Hadoop and MapReduce or non-incremental, i.e., they recalculate the result at the arrival of new data. In this paper we propose an incremental and distributed method of reasoning over very large OWL ontologies based on SPARK, which offers a reduced execution time as it loads the RDF triplets in memory and not in disk. Our method allows creating transfer inference forest (TIF) and effective assertional triples (EAT) to reduce disk space and simplify and accelerate the reasoning process.

Keywords: semantic web; ontology reasoning; ontology web language; OWL; OWL Horst; semantic web rule language; SWRL; SPARQL; Hadoop; SPARK.

DOI: 10.1504/IJCC.2019.10021217

International Journal of Cloud Computing, 2019 Vol.8 No.2, pp.140 - 149

Received: 01 Feb 2018
Accepted: 09 Nov 2018

Published online: 10 May 2019 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article