PAA海量数据上一种有效的近似聚集查询算法.pdf

计算机研究与发展 of Researchand 51(1)：41—53，2014 JournalComputer Development PAA：海量数据上一种有效的近似聚集查询算法 韩希先 李建中 高 宏 (哈尔滨工业大学计算机科学与技术学院 哈尔滨 150001) (hanXiXian@) Efficient onⅣIassiVeData PAA：An AggregatiOnAlgorithm Approximate Han Gao Xixian，Li Jianzhong，andHong Scie，lceⅡnd 150001) (Schoo￡o，Co，np“￡er T、ec^no￡093，，HⅡr6in工ns￡i￡M￡eo，’rec^扎。厶)gy，Hnr6in Abstract isa usedbut indatabase Aggregationcommonly time—consumingoperation systems． Relative toexact isoftenmoreattractivetoreturnan resultwiththe Comparedquery，it approximate that errorboundtouserinamuchfaster time．However，wefind noneofthe required response methodscan onmassiVedatawith and approximateaggregation arbitraryaccuracy previous process Anovel PAAis to highefficiency． algorithmproposedef“cientlyprocessapproximateaggregation isdividedinto withan confidenceinterval．Thedata ofdimensionalattributes arbitrary space multiple ofthesamecubesize．Each maintainsthe whosedimensionalattributes hypercubes partition tuples faUintothe random RSis ontable．PAAconsists correspondinghypercube．Asamplepre—constructed If RSin 1 doesnot the oftwo the resultobtained confidence stages． approximate by stage satisfy