Cloud Computing Environment Gml Space-time Clustering Data Mining Research

Lots of implicit knowledge and information are exist in GML spatio-temporal data, including spatio-temporal model and its characteristic,the general relationship between spatio-temporal dataand common data, and the general data characteristicexists in GML spatio-temporal data and so on. Human can learn about the the natural worldby means ofgetting relation, disciplineand interaction existing in nature,also these are used to making decision for our production and life. However, due to the GML spatio-temporal’s temporal-spatial and half structural characteristic,it can’t use accurate mode to define GML spatio-temporal data. So it si much more complicated to extract information from GML spatio-temporal datathan traditional data. At the same time, because of characteristics of quantity and the various computing intensive, these limit the information processing progress some dgree.To resolve all of these problem,this paperput forward two parallel clustering mining algorithmin parallel computing Hadoop environment.In this paper the auther design a parallel clustering GML mining prototype system,and At last,Showing the clustering visualizesto the form of map.(1)This paper gives two kinds of parallel clustering algorithm for mining GML spatio-temporal series data.The first kind is put forward with K-means clustering mining algorithm based on time sequence of GML spatio-temporal similarity, Considering the spatial attribute and temporal series together to metric the similaritybetween the two spatial adjacent objects.Then through the K-means clustering algorithm conduct the mining.(2)The second is according to the definition of spatial neighborhood to get the GML spatial neighborhood in GML spatio-temporal objects, and then in this neighborhood meature the two objects of temporal series of the attribute based on temporal sequence to getting the similarity, combining parallel DBSCAN (STN_PDBSCAN) clustering algorithm for data mining.(3)Through constructing Hadoop distributed parallel computing platform, using MapReduce programming model to realize the two kind of clustering algorithmK-means and STN_PDBSCAN parallel process. By designing and implementing a parallel GML spatio-temporal clustering prototype mining system, can running the two parallel clustering algorithm using GML format attributes of the meteorological data.Through the experimental results we verify effectiveness of the twoparallel algorithm,and quality of high performance.Good performance of the algorithm can be expanded.(4)Finally, we show the results of cluster income in the form of map.