Research And Application Of GIS-based Spatial Data Mining Technology

Spatial data mining (SDM) refers to extracting interesting rules from spatial database,such as spatial patterns and characteristics, the universal relations of spatial and non-spatialdata and other intrinsic characteristics behind spatial data. It is the most active branch ofdatabase technology and knowledge acquisition method. Geographic Information System isa system based in spatial database, supported by computer software and hardware, usingsystem engineering and information science theories to manage and analyze these spatialgeographic data, with the purpose of providing information for planning, management,decision-making and research.There is an abundant amount of data in GIS database, and the main function of recentGIS system is to acquire, query and analyze data, while as refer to tracing the potentialrelationship and regulation among data, it is just limited. To solve this problem, theintroduction of spatial data mining to GIS and effective integration with GIS functionalitymake it possible to find out the intrinsic knowledge and regulation behind the large volumespatial data hidden in GIS database.The main research content of this dissertation is to discover knowledge and interestingregulation from GIS spatial database through SDM. Firstly, the dissertation introduces thedefinition and characteristic of GIS, with a detailed presentation of the function and spatialdata structure and hence the development trend of GIS. Secondly, it introduces SDM asregard to its concept definition, architecture, mining process and theoretical approach, andalso analyzes its main application. Then, the emphasis is laid on mining association rules,i.e. Apriori algorithm and its improvement. In consideration with the characteristic ofspatial data, a spatial association rule algorithm based on Concept Hierarchy Tree is putforward. Finally, a detailed research is made on integration mode of SDM and GIS.AGIS-Based Spatial Data Mining System is presented, and its effectiveness is also evaluatedthrough experiments.