Conceptual Framework for Modeling Dynamic Paths from Natural Language Expressions

122467470By Kathleen Stewart Hornsby and Naicong Li.
Transactions in GIS, Volume 13 Issue s1, Pages 27 – 45

Abstract: Text documents frequently contain descriptions of different kinds of movements by individual persons, groups, animals, vehicles, or other moving objects. Comprehending and modeling the semantics of movement is an area of interest for geographic information science. In this article, we show how text documents that contain movement verbs can be analyzed for deriving representations of movement or dynamic paths. A conceptual framework is presented that provides the foundations necessary for deriving dynamic paths automatically from natural language descriptions and representing these dynamic paths in an information system, such as a geographic information system. In this research, a linguistic analysis of dynamic paths is presented and linked to a spatiotemporal representation of paths. We show how movement descriptions in text can be mapped to a set of elemental components including source, destination, route, direction, distance, start time, end time, and duration. Together, this set of path components captures the spatiotemporal characteristics of the path of a moving object as described using natural language. A systematic examination of these components builds a foundation for understanding more complex scenarios involving discourse (composed of consecutive sentences). Additional aspects reflecting important semantics about the movement characteristics of objects and discussed here are the shape of the path and granularity of modeling.