Path reconstruction based on natural language and the uncertainty problems

It is an important research issue to build systems supporting natural language comprehension in social geographic information service. In this paper, representation of a path in natural language (NLRP) is discussed. Based on the syntax model of NLRP, an algorithm named PRA (path reconstruction algorithm) used for generating linear path according to the reference map and NLRP expression is described. Inevitably, a NLRP expression contains some uncertainties, which mainly come from the spatial cognition process and imprecise expression. The uncertainties of NLRP include uncertainties of direction, distance and topologic relationship. They make the NLRP expression ambiguous and bring some problems to PRA. The solution is also discussed