Probabilistic Information Integration and Retrieval in the Semantic Web

The Semantic Web (SW) has been envisioned to enable software tools or Web Services, respectively, to process information provided on the Web automatically. For this purpose, languages for representing the semantics of data by means of ontologies have been proposed such as RDF(S) and OWL. While the semantics of RDF(S) requires a non-standard model-theory that goes beyond first order logics, OWL is intended to model subsets of first order logics. OWL consists of three variants that are layered on each other. The less expressive variants OWL-Light and OWL-DL correspond to the Description Logics SHIF(D) and SHOIN (D) [1], respectively, and thus to subsets of First Order Logics [2]. While RDF and OWL are W3C recommendations and hence a kind of standard, a lot of proposals emerged recently for representing Logic Programming (LP) variants on the Web. Such proposals are e.g. SWRL and WRL. Furthermore, a working group exists at the W3C for defining a rule interchange format. Therefore, it can be expected that rule languages will play an important role in the SW. The Description Logics (DL) and the LP paradigm are orthogonal having just a small subset in common [3] and a comparison reveals a balanced amount of advantages and disadvantages of one compared to the other e.g. concerning the efficience of certain reasoning tasks [4]. The SW will consist of independent peers each providing information that describes overlapping domains by different ontologies or logic programs specified in different knowledge representation languages of the DL and the LP paradigm. In order to enable intelligent software tools to utilize the information represented by these peers in a coherent manner, the ontologies and logic programs need to be aligned by means of mappings. An number of approaches for learning mappings between ontologies exist already [5]. Most of them detect very simple matchings and can be used for learning mappings between ontologies and logic programs as well.