Comparing Representations of Geographic Knowledge Expressed as Conceptual Graphs

Conceptual Graphs are a very powerful knowledge and meaning representation formalism grounded on deep philosophical, linguistic and object oriented principles [1], [2]. Concerning geographic knowledge representation and matching, the study and analysis of geographic concept definitions plays an important role in deriving systematic knowledge about concepts and comparing geographic categories in order to identify similarities and heterogeneities [4]. Based on the proposed algorithm for the representation of geographic knowledge using conceptual graphs, we also present a method that takes into consideration the special structure of conceptual graphs and produces an output that shows how much similar two geographic concepts are and hence which concept is semantically closer to another. For producing the conceptual graph representation of any geographic concept definition we follow two steps, tagging and parsing, while for measuring the similarity between two geographic ontologies we apply proper modifications to the Dice coefficient that is mainly used for comparing binary structures.