Reasoning With Provenance, Trust and all that other Meta Knowlege in OWL

For many tasks, such as the integration of knowledge bases in the semantic web, one must not only handle the knowledge itself, but also characterizations of this knowledge, e.g.: (i) where did a knowledge item come from (i.e. provenance), (ii) what level of trust can be assigned to a knowledge item, or (iii) what degree of certainty is associated with it. We refer to all such kinds of characterizations as meta knowledge. Approaches for providing meta knowledge for query answers in relational databases and RDF repositories, based on algebraic operations, exist. As query answering in description logics in general does not boil down to algebraic evaluation of tree shaped query models, these formalizations do not easily carry over. In this paper we propose a formalization of meta knowledge, which is still algebraic, but allows for the computation of meta knowledge of inferred knowledge in description logics, including reasoning with conflicting and incomplete meta knowledge. We use pinpointing to come up with meta knowledge formulas for description logics, which then can be evaluated algebraically. We describe and evaluate our prototypical implementation.