4s-reasoner: RDFS Backward Chained Reasoning Support in 4store

Abstract

This paper describes the design and implementation of backward chained clustered RDFS reasoning in 4store. The system presented, called "4s-reasoner", adds no overhead to the import phase and yet performs reasonably well at the query phase. We also demonstrate that our solution scales over clusters of commodity servers providing an optimal solution that balances infrastructure cost and performance over tested data sets with up to 500M triples. In addition we have shared our implementation under GNU license and a first release is available to be used by the community.

Storing massive Resource Description Framework (RDF) data: a survey

“…So, another major open issue is the incorporation of RDFS (RDF Schema) and OWL (Web Ontology Language) into RDF data management for reasoning. Currently, relatively little work has been done on the impact of reasoning on RDF data management (Theoharis et al, 2005;Salvadores et al, 2010Salvadores et al, , 2011Bishop et al, 2012;Sperka & Smrz, 2012). Stardog 37 as well as Oracle RDF Store 38 have support for the most advanced OWL profiles, but their scalability is questionable.…”

The Design and Implementation of Minimal RDFS Backward Reasoning in 4store

“…In [10] we presented the logical model for bind', a preliminary work that did not take into account the Minimal RDFS fragment and was built upon a subset of semantics from [5].…”

“…These datasets progressively grow from 13M triples -LUBM(100,0) -triples to 138M triples LUBM(1000,0). In [10] we presented a preliminary benchmark that demonstrates that 4sr can handle SPARQL queries with up to 500M triple datasets; this benchmark shows the overall performance of the whole system. The type of benchmark we analyse in this paper, instead of studying performance for big datasets, studies how the bind operation behaves when trying to find solutions that require Minimal RDFS reasoning under different conditions, i.e.…”

Κατανεμημένη Αποτίμηση Επερωτήσεων Και Συλλογιστική Για Το Μοντέλο RDF Σε Δίκτυα Ομότιμων Κόμβων

“…In [126], the authors present the reasoning engine of 4store [53] which runs in a backward chaining fashion. 4store is a clustered RDF store which uses the subject of each triple to decide to which cluster node the triple should be stored.…”

“…The reasoning engine of 4store works in a backward chaining fashion but keeps all RDFS information at one cluster node. The experiments presented in [126] were conducted in 5 Dell PowerEdge R410 machines, each of them with 4 dual core processors at 2.27GHz, 48GB memory and 15k rpm disks managing to scale to 500 million triples.…”