Loris Bozzato

CKR: a general framework for context in Semantic Web

The Center for Computer Science (CCS) and the Wolfgang Pauli Institut (WPI) will host a talk by Loris Bozzato.

DATE:Monday, February 25, 2013
TIME:11:00 c.t.
VENUE:Seminar room Goedel (Favoritenstrasse 9-11, ground floor, access through courtyard)

ABSTRACT

In this talk we present the latest results and ongoing work on the new formalization and implementation for the CKR (Contextualized Knowledge Repository) contextual framework. In this new realization of the framework, different aspects of contextualization have been redefined and extended starting from a study on the representation needs for contexts in Semantic Web applications.

We will first introduce the new DL based formalization of CKR: intuitively, a CKR is composed by a global context, a knowledge base describing both the contextual structure and the globally valid object knowledge, and a set of (local) contexts, which contain local object knowledge possibly with references to the knowledge of other set of contexts. Syntax and semantics of global and local contexts are based on a restriction of the SROIQ DL to the form of OWL RL axioms. Inference in CKR is obtained by forward rules: we formalize the forward closure process through the definition of a materialization calculus for instance checking in CKR.

After introducing these theoretical definitions for CKR, we present its realization in a prototype based on SPRINGLES (SParql-based Rule Inference over Named Graphs Layer Extending Sesame), a system we are developing to enable inference over multiple RDF named graphs in a Sesame-compliant triple store. We will present how CKR inference rules are encoded and their execution strategy specified in SPRINGLES and we will show a brief demo of the prototype.

We conclude our talk by presenting the future directions for extending CKR to Answer Set Programming tools: the combination will permit to include in our framework some form of online reasoning and. possibly to scale over large and evolving datasets.

 

CONTACT

With kind support of the Center for Computer Science (CCS) and the Wolfgang Pauli Institut (WPI).

Comments are closed.