Summary | The representation of temporal and spatial information and reasoning about such information are important subproblems in many applications, such as geographical information systems (GIS), natural language understanding, planning, diagnosis of technical systems, and document interpretation. This information is often only available qualitatively, for instance when a natural language sentence or a GIS query has to be interpreted. In this case, qualitative approaches to reasoning about space and time have to be used. For these aims, a number of different qualitative temporal and spatial calculi have been developed. However, only a few of these approaches address the combination of temporal and spatial representation and reasoning. If a GIS query also addresses the temporal dimension or if the temporal development of spatial configurations is to be described, such a combination of temporal and spatial qualitative information is obviously necessary. The proposed project aims to develop such qualitative spatio-temporal calculi including an analysis of the logical properties and their decidability and computational complexity. In addition, it is also intended to develop algorithms and to provide prototype implementations of reasoning systems. In the project, we will mainly concentrate on constraint-based reasoning, i.e., we will analyze the process of finding feasible assignments of the variables. However, we will also consider the problem of matching a qualitative description to a given scenario. The computational investigations will thus be complemented by psychological studies to investigate the cognitive adequacy of formal calculi, i.e., to which degree a set of formally defined relations corresponds to the mental set of relations when individuals conceptualize situations in space and time. In the suggested empirical investigations we will use the grouping task paradigm, which is one of the best-known methods to investigate conceptual knowledge in psychology. The main idea is that spatial arrangements are assessed as similar if they are instances of the same relational concept, or assessed as dissimilar if they are instances of different concepts. This psychological concept is closely related to the conceptual neighborhood theory in AI. Overall, the project is intended to combine computational and psychological results concerning spatial calculi on a level that has not been reached so far.
|