A qualitative spatial reasoning approach in knowledge discovery in spatial databases

Abstract

Spatial data mining refers to the extraction of implicit knowledge, spatial relationships, or others interesting patterns not explicitly stored in spatial databases. Spatial databases like geographic databases in geographic information systems usually store large amounts of spatial information. Queries in these systems involve the derivation of different spatial relations that are not explicitly stored. Qualitative spatial reasoning has been proposed as a complementary mechanism for the automatic derivation of these relations. Qualitative spatial relations are specified using of a small set of symbols, like North, close, etc., and are manipulated through a set of inference rules. In this paper is presented a new approach to the process of knowledge discovery in spatial databases, where geographic identifiers give the positional aspects of geographic data. These identifiers are manipulated using qualitative reasoning principles, allowing the inference of new spatial relations required in the knowledge discovery process. The exploration, with this approach, of a demographic database allowed the discovery of implicit spatial relations that characterise the analysed geographic and demographic data. The overall system, including the spatial reasoning mechanisms and the discovery process, was implemented in Clementine. Between others facilities, the system allows the integration of the relevant discoveries in a database of patterns, manipulated by a geographic information system, enabling its visualisation in a map.