Database Architecture for Specifying and Modeling Spatio-Temporal Relations

Abstract

The applications of systems for spatiotemporal relations representation and reasoning range from natural language processing to computer vision and robotics. Our most recent research is concerned with a problem of specifying and modeling spatiotemporal and cause-effect relations between objects in a real physical environment. We propose a method based on defining environmental database as a network (graph) model which originates from the object-attribute (OA) dataflow computing architecture and is similar to a frame knowledge representation scheme. In this paper, we propose a pseudo-physical database architecture which integrates spatiotemporal aspects of the application domain and provides support for reasoning with an OA-grammar of the graph transformation system, a proprietary calculus for spatial reasoning, and an OA-language. A prototype of the OA-DB has been implemented and used to model an application domain-Natural Language Processing.