Models for Conceptual Geographical Prepositions Based on Web Resource

If there exists a mismatch between the way people use prepositional phrases and a computer embedded algorithm that handles those phrases, users cannot get proper results for their queries. This causes serious problems in decision making processes. In this paper, such problems are discussed for the case of Geographical Information Systems (GISs). In order to obtain human subjective cognition of the world, we used text data on the Web. We extracted common geographical prepositions such as “near”, “between”, and “in front of” out of a large number of web pages, and investigated their usage. The results show that the models supported by currently used computer algorithms are very different from people’s actual usage of geographical prepositions. This paper analyzes why such mismatches occur. Results show that the main factors affecting the meaning of geographical preposition are landmark characteristics and interactions. We propose a basic method to overcome such