SOWL: a framework for handling spatio-temporal information in OWL

Abstract

The Semantic Web extends existing World Wide Web (WWW) with formal machine readable semantics that enable better understanding of Web content by both, people and machines. The OWL ontology language forms the basis of Semantic Web technologies by providing the means for defining formal semantics of concepts and their prop- erties. Semantic Web standards also provide with the formalism and mechanisms for supporting reasoning and querying of information in OWL. Representing dynamic information for concepts evolving in time and space is an important issue to deal with. This calls for mech- anisms for representing dynamic properties of objects. In addition, these mechanisms must over support for qualitative information (i.e., information represented using natural language expressions such as \before" for time or \north" for space) which are common in natural language expressions in many applications. The current thesis addresses all these issues: representation of dynamic concepts is achieved using the \4D- uents" or, alternatively, the \N-ary relations" mechanisms; Both mechanisms are thoroughly explored and are expanded for representing qualitative (in addition to quantitative) spatio-temporal information in OWL. Temporal repre- sentations based both, on intervals and on time instants (points) are supported and compared. Our approach models spatial content in description logics using the eight relations of the Region Connection Calculus (RCC8) complemented by directional relations encoded by cone-shaped or, alternatively, projection-based relations. Qualitative spatial and temporal information is supported us- ing sets of SWRL rules and OWL axioms overing a sound, complete and tractable reasoning procedure based on path consistency apply- ing on the supported sets of relations. Polynomial time complexity of spatio-temporal reasoning is achieved by restricting the supported sets of relations to \tractable" sets. Furthermore, rules enforcing cardinal- ity constrains on dynamic properties are defined. Reasoning support for qualitative information and restriction checking is also introduced in this work for increasing expressive power of existing approaches. Building-upon existing Semantic Web standards and on the idea of integrating reasoning support into the proposed representation are important advantages of the proposed approach. Query support in SOWL is realized with two languages referred to as TOQL and SOWL Query Language. Both, are high-level query languages, independent from the underlying SOWL representation so that, the user need not be familiar with the peculiarities of the onto- logical spatio-temporal representation in SOWL (i.e., the 4D- uents or the N-ary approach). TOQL handles ontologies almost like rela- tional databases and relies on the idea of extending SQL with spatio- temporal operators. Respectively, SOWL builds-upon SPARQL, the current standard of the Semantic Web, which is also extended with a set of spatio-temporal operators similar to those introduced in TOQL.