Mapping the knowledge of spatial data mining

Abstract

Map is the model of spatial data. It is the basis of spatial database. Because of the complexity of spatial data mining and the diversity of spatial relations, too many spatial rules have been discovered in spatial data mining. In order to comprehending the knowledge contained in spatial rules, spatial data mining visualization has attracted a great deal of attention. In fact, map is not only one of the objects of spatial data mining, but also one of the predominant visualization tools of spatial rules or knowledge. This paper proposes that map is the proper visual method of spatial rules from spatial data mining. By the visual variables of map expressing, such as static variables, including shape, size, color, brightness, design, texture and dynamic variables, including video, sound, and so on, Map can visualize static and dynamic knowledge. From the point of view of map expressing, spatial knowledge or rules can be classified four categories, spatial characteristic rules, spatial distribution rules, spatial relation rules and temporal-spatial evolution rules. We propose visualization models and methods for each category by virtue of the visual variables of map expressing. Along with the development of 3D map, map can visualize spatial knowledge better and facilitate comprehending the spatial rules from spatial data mining.