Abstract
Owing to the GIS technology, a vast volume ofspatial data has been accumulated, therebyincurring the necessity of spatial data miningtechniques. In this paper, we propose a new spatialdata mining system named SD-Miner. SD-Minerconsists of three parts: a graphical user interface forinputs and outputs, a data mining module thatprocesses spatial data mining functionalities, a datastorage model that stores and manages spatial aswell as non-spatial data by using a DBMS. Inparticular, the data mining module provides majorspatial data mining functionalities such as spatialclustering, spatial classification, spatialcharacterization, and spatio-temporal associationrule mining. SD-Miner has its own characteristics1) It supports users to perform non-spatial datamining functionalities as well as spatial data miningfunctionalities intuitively and effectively; (2) Itprovides users with spatial data mining functions asa form of libraries, thereby making applicationsconveniently use those functions. (3) It inputsparameters for mining as a form of database tablesto increase flexibility.
Keywords: Spatial data, Spatial Data Mining, SDMiner
1 Introduction
Due to the development of information technology,a vast volume of data is accumulated on manyfields. Since automated methods for filtering/analyzing the data and also explaining the resultsare required, a variety of data mining techniquesfinding new knowledge by discovering hidden rulesfrom vast amount of data are developed [5].In the field of geography, due to the developmentof technology for remote sensing, monitoring,geographical information systems, and globalpositioning systems, a vast volume of spatial data isaccumulated. Also, there have been many studies ofdiscovering meaningful knowledge from the spatialdata.Since the spatial data has its own characteristicsdifferent from the non-spatial data, direct using ofgeneral data mining techniques incurs manydifficulties [9]. So there have been many studies ofspatial data mining techniques considering thecharacteristics of the spatial data [1-3, 6-10].However, commercial tools for spatial data mininghave not been provided. Currently, manycommercial data mining tools are available, butthese tools support the spatial data miningfunctionalities. Also, while some academic spatialdata mining tools such as GeoMiner [4] areavailable, there are almost no commercial spatialdata mining tools. So, for easy using of spatial datamining for real spatial data applications,developments of spatial data mining tools areneeded.In this paper, we propose a new spatial data miningsystem named SD-Miner. SD-Miner supports fourimportant spatial data mining functionalities: spatialclustering [10], spatial classification [8], spatialcharacterization [1], and spatio-temporalassociation rule mining [7]. We first analyzecharacteristics of previous spatial data miningtechniques and suggest techniques to improve theirefficiency in developing SD-Miner.The rest of this paper is organized as follows.Section 2 explains four important spatial datamining functionalities adopted in SD-Miner.Section 3 introduces the structure of SD-Miner.Section 4 discusses some our enhancements forimproving the efficiency of SD-Miner. Section 5gives the final conclusion of this paper and thefuture work.
2 Spatial data mining
In this section, the concept and characteristics ofspatial data mining techniques adopted in SDMinerare explained. Spatial data mining considersthe characteristics of the spatial data in data miningand thus can be regarded as a spatial extension ofdata mining. i.e., spatial data mining discovershidden interesting spatial relationships, variousspatial patterns, and knowledge from a spatialdatabase [5]. Spatial objects are composed of notonly general attributes represented as text but alsospatial attributes such as point, line, and surface oftwo- or three-dimensional space [3]. The spatialobjects also have topological information. Thespatial data mining system must have functions forextracting of patterns of spatial and non-spatialattributes.


Download full report
http://ieeexplore.ieeeiel5/5353577/53607...er=5360924