Describing the topological relationships between 2-D objects using natural language
Space plays a fundamental role in human cognition and spatial relationships often act as an organizing structure for further conceptual material. A computer system capable of describing, using natural language, the spatial relationships between objects could greatly enhance many existing applications, including Geographic Information Systems. This thesis proposes a system for generating linguistic descriptions of the spatial relationships between 2-D objects. The objects considered are not necessarily convex, nor connected, and they may have holes in them. The descriptions convey the primary topological relationships and use directional refinements to indicate where these relationships hold true. The proposed system relies on the computation of Allen F-histograms and the fuzzified set relations DISJOINT, OVERLAPS, SUBSET (CONTAINED BY), SUPERSET (CONTAINS), and EQUALS. The system is tested using sets of synthetic and real data.