Natural-language representation and query of linear geographic objects in GIS
People usually use qualitative terms to express spatial relations, while current geographic information systems (GIS) all use quantitative description to store spatial information. The abilities of current GIS to represent spatial information about geographic space are limited, and it is inconvenient for GIS users without professional training. In this dissertation, three attempts are made to conquer the limitation of current GIS so as to implement an intelligent and human-friendly GIS. First, a human-subjects test is conducted to find out how natural-language descriptions of spatial relations of linear objects are determined by the geometric configurations of the objects. The results indicate that topology and metric properties decide people's agreement to the spatial predicates, and they have different degrees of importance for the various spatial predicates. Second, the result of the human subjects study is used to analyze the relationship between the natural-language terms and the geometric and topologic relations between the objects. A series of metric indices, such as angles, distance ratios, splitting ratios, and overlap ratios, are measured. The evaluations given by the human subjects are changed to membership values which represent how strongly the spatial relations of two liner objects belong to the groups described by the specific natural language terms. A decision tree algorithm is applied to formalize the spatial predicates with metric indices and topologic value of the 9-intersection. Finally, based on the formalization of the natural-language terms, a natural-language interface of ArcMap that can query spatial relations is implemented with SNePS (the semantic network processing system), a natural language representation system. In this interface, SNePS accepts users' queries in natural language, and translates them to a format that is understandable by computer systems. Then ArcMap gets the translated queries, responds to them, and displays the result to the users. The rules generated from the decision trees are used in ArcMap to query the natural-language spatial relations.