Abstract: Semantic similarity plays an important role in the knowledge sharing and data integration. Currently, in some geographical information related applications, the similarity between geographical concepts was measured with the semantic distance of concepts within domain ontology, which was usually captured from the specified concept taxonomies directly. By this method, though the similarity between concepts can be defined easily and fast, because of the alteration of the concept taxonomy, the discrepancies of similarity may exist even in a same concept pair and possibly cause the mistakes in the similarity measurements. This article proposes a semantic similarity measurement model based on the intrinsic features of the concepts, represented by the attribution enumeration for the need of fundamental geographical domain application. In this model, a set of ontology properties, representing the natural features of each concept, were defined to model each fundamental geographic concept. After that, the similarity of each pairs of the concepts can be measured by integrating the similarity of each ontological property and its corresponding weight, which were defined by the experts. In this paper, a hundred samples of the concepts of the fundamental geographic information were chosen to compute their similarity. The results showed that this proposed approach achieved good performance in measuring the similarity of the fundamental geographic information concepts.

Key words: ontological property, semantic similarity, attribution enumeration, fundamental geographic information concept