Introducing Context into RDF Knowledge Bases.

Abstract

Large-scale knowledge representation (KR) with RDF unveils shortcomings which become obvious when facts have to be further qualified with contextual aspects to represent anything else but simplistic binary predicates. Motivated by requirements in the VIKEF project, in which we are required to store a large amount of information extracted from a heterogeneous set of documents and encoded in RDF triples, we are proposing an architecture for modelling context in RDF knowledge bases. Our approach -- based on well-researched theories of context in KR -- avoids issues that other approaches face, by preserving standard RDF within a context and adding context semantics and relations between contexts around the standard. In this paper we will present our approach, including formal definitions of our extensions and an illustration of further works.