Representing Contexts and Exceptions in Semantic Web Data with CKR
Sprache des Titels:
Representation of context dependent knowledge in the Semantic Web has been recognized as a relevant issue: as a consequence, a number of logic based formalisms have been proposed in this regard. In our recent works, in response to this need, we presented the description logic-based Contextual-ized Knowledge Repository (CKR) framework.
A CKR knowledge base has a two layered structure, modelled by a global context and a set of local contexts: the global context contains the meta knowledge defining the properties of local contexts, but also holds the global (context independent) object knowledge that is shared by all of the local con-texts; the local contexts contain context-dependent knowledge and, possibly, references to the knowledge of other (classes of) contexts.
In this talk we present the latest results and ongoing work on the formalization and implementation for the CKR framework. We will first introduce the descrip-tion logic definition of the framework and an inference procedure defined in terms of a datalog based materialization calculus. Then, we will show how the framework has been implemented over RDF data and the reasoning procedure has been realized using SPARQL based rules.
We will then present our current work in extending the framework with a notion of non-monotonic justifiable exceptions. Intuitively, we extend CKR definitions with the possibility to represent global defeasible axioms: such axioms hold in the local contexts only for the instances for which it does not exists a justification for their overriding. Over such semantics, we will present a translation of extended CKRs to datalog programs with negation under answer sets semantics (as an extension of the original materialization calculus) and we provide an overview of its prototype implementation in CKRew (CKR datalog rewriter).