一种基于主存_tree的高维数据自相似连接处理.pdfVIP

ISSN 1000- 1239PCN 11- 1777PTP Journal of Computer Research and De elopment 46( 6) : 995- 1002, 2009 $-tree 1, 3 1, 2 刘 艳 郝忠孝 1 ( 150080) 2 ( 15000 1) 3 ( 130022) ( liuy an6374@ y ahoo. com. cn) A $- Tree Based Similarity Join Processing for High-Dimensional Data 1, 3 1, 2 Liu Yan , and Hao Zhongxiao 1 ( Colleg e of Comp uter and Control, H arbin University of S cience and Technology , H arbin 150080) 2 ( Colleg e of Comp uter Science and Technolog y, H arbin Institute of Technology , H arbin 15000 1) 3 ( Colleg e of Comp uter Science and Technolog y, Changchun Univ ersity , Changchun 130022) Abstract T he similar ity join, an important data mining pr imiti e, can be successfully applied to speeding up applications such as similar ity sear ch, data analysis and data mining. So far most of researches focus on the ex ecution of high-dimensional joins o er large amounts of disk based data. T he increasing sizes of main memory a ailable on curr ent computers, and the need for efficient processing of spatial joins suggest that spatial joins for a large class of problems can be processed in main memory. $- tr ee is a no el mult-i le el index structure, it can speed up the high-dimensional query in main memory en ironment and has been pr o en to be an efficient index method. Each le el in the $- tree represents the data space at different dimensionalities: the number of dimensions increases tow ards the leaf le el w hich contains the data at their full dimensions. T he remaining dimensions ar e obtained using principal component analy sis. U sing the properties of $-tr ee, a simila