Materialization Calculus for Contexts in the Semantic Web

Representation of context dependent knowledge in the Semantic Web is becoming a recognized issue and a number of DL-based formalisms have been proposed in this regard: among them, in our previous works we introduced the Contextualized Knowledge Repository (CKR) framework. In CKR, contexts are organized hierarchically according to a broader-narrower relation and knowledge propagation across contexts is limited among contexts hierarchically related. In several applications, however, this structure is too restrictive, as they demand for a more flexible and scalable framework for representing and reasoning about contextual knowledge. In this work we present an evolution of the original CKR (based on OWL RL), where contexts can be organized in any graph based structure (declared as a metaknowledge base) and knowledge propagation is allowed among any pair of contexts via a new ”evaluate-in-context” operator. In particular, we detail a materialization calculus for reasoning over the revised CKR framework and prove its soundness and completeness. Moreover, we outline the current implementation of the calculus on top of SPRINGLES, an extension of standard RDF triple stores for representing and rule-based inferencing over multiple RDF named graphs.