Efficient Utilization of DBMS Potential in Spatial Data Mining Applications - Neighborhood Relation Modeling Approach

摘要

In the core concept of spatial data mining we need to investigate the neighbors of many objects in the single run of typical data mining algorithm. This means that in spatial data mining algorithm we have to efficiently process the neighborhood relations. An integration of spatial data mining algorithms and the potential of spatial database management system (SDBMS) will help efficiently providing general concept of neighborhood relation and its implementation. If properly integrated, this will speed up the industrial application development of spatial database and applying the traditional data mining concept on that[1][4][5][6]. Here in this paper the neighborhood relation and neighbourhood index has been discussed and the related small set of database primitives has been defined. The proposed basic operations reduce the search space and support well the typical data mining algorithms. Such database primitives have been implemented on top of a commercially available Spatial Database Management System (SDBMS) Oracle10g [6][7].