Spatial Database, Spatial Data Mining Research

The coming of the information era makes knowledge become an important factor for decision-making, but indigent access means of knowledge restricting the comprehensive utilization of the data. GIS technology wider used, data mining, collection technology of spatial data, spatial database development and maturity, and the scarcity of analyzing function in GIS, above-mentioned spur appear knowledge found from spatial database—Spatial Data Mining. Among them, the co-location mining is a novel spatial data mining methods.Co-location pattern represents the instances of spatial feature space frequent association .At present the difficulties of Mining spatial co-location pattern is: complex spatial relationships of spatial instances in continuous space. Therefore, most computing time used to separate instances from co-location pattern..But, in some applications do not need to identify all the co-location pattern, only some pattern about the attributes.For this situation, the paper proposes a novel algorithm for co-location mining .This algorithm use relations between the neighbouring to find co-location as the basic method, and taking into account the negative co-location rules. In order to reduce calculation work ,it use star neighborhood computing and define interest pattern to delete redundancy model. Simulation experiments show that the algorithm can find positive co-location rules and negative co-location rules at the same time . It improve the efficiency of co-locationmining and accuracy of the results.