Towards Spatial Reasoning in the Semantic Web: A Hybrid Knowledge Representation System Architecture

Abstract

Abstract. Environmental databases store a wide variety of data from heterogeneous sources which are described with domain-specific terminologies and refer to distinct locations. In order to make them accessible also to non-expert users, terminological concepts and spatial relations must be represented in a way that they can be exploited for searches. In this paper we propose a hybrid knowledge representation system architecture which integrates terminological and spatial aspects of the application domain and provides support for reasoning with RCC, a well-known calculus for spatial reasoning, and the Semantic Web's ontology language OWL. Our approach is motivated by the observation that RCC cannot be expressed in OWL without a major revision of the latter. Issues of upholding consistency of the knowledge base in view of an evolving ontology and of computational complexity are discussed.

Improving an RCC-Derived Geospatial Approximation by OWL Axioms

“…The relations between individual regions can then be asserted in terms of RCC-5 in the ABox of a knowledge base. Based on this, a full-fledged spatio-terminological reasoning service as outlined in [7,8] can be developed.…”

“…In this model, eight out of nine relations can be interpreted in the same way as we interpret the RCC-8 relations, namely as spatial relations between polygons in the integral plane [7]. Only the ninth relation is specific for the model.…”

“…It rather adds to the knowledge about the combination of quantitative numerical approaches with qualitative symbolic (i.e., logic) approaches in general. The presented approach is related to previous work on combining RCC with OWL [7,8] by addressing the calculation of a geospatial world description for assertion in the ABox of a knowledge base. Reasoning with both RCC and OWL will make use of the asserted world description.…”

End-User Assisted Ontology Evolution in Uncertain Domains

“…An open search was realized by an expansion scheme [14] and the reasoning is based on DL. Further research is performed with the objective to extend the knowledge base by a so-called RCCBox representing composition tables for spatial inference based on the Region Connection Calculus (RCC) [15,16]. In late 2008, a first prototype shall be released for a test cycle of selected expert users at the Swiss Federal Office for the Enviroment.…”

A fuzzy extension of the semantic Building Information Model

“…Notice that this would require to calculate pair-wise distances between all building elements, which may be computationally expensive and require additional optimizations. Other interesting improvement can be the combination with RCC (Region Connection Calculus) predicates, which allow symbolic rep-695 resentation and reasoning with topological relations [42]. RCC is not directly supported in crisp OWL, but still, it is possible to create axioms to model common relations, like tangent, overlap, disjoint, etc., and instantiate them by relying on an external module [43].…”