• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2012, Vol. 34 ›› Issue (6): 59-64.

• 论文 • 上一篇    下一篇

一种基于逆支配点集的数据流Top-k计算方法

甘亮1,2,于莉莉3,李润恒1,贾焰1,金鑫4   

  1. (1.国防科学技术大学计算机学院,湖南 长沙 410073;2.第二炮兵指挥学院三系,湖北 武汉 430012;
    3.北京航空航天大学计算机学院,北京 100191;4.长沙民政职业技术学院软件学院,湖南 长沙 410004)
  • 收稿日期:2010-07-20 修回日期:2011-02-15 出版日期:2012-06-25 发布日期:2012-06-25
  • 基金资助:

    国家863计划资助项目(2006AA01Z451,2007AA01Z474)

A Reverse Dominant Point Set Based Algorithm for Top- k  Queries in DSMS

GAN Liang1,2,YU Lili3,LI Runheng1,JIA Yan1,JIN Xin4   

  1.  (1.School of Computer Science,National University of Defense Technology,Changsha 410073;
    2.Department Three,The Second Artillery Command College,Wuhan 430012;
    3.School of Computer Science and Engineering,Beihang University,Beijing 100191;
    4.Changsha Social Work College,Changsha  410004,China)
  • Received:2010-07-20 Revised:2011-02-15 Online:2012-06-25 Published:2012-06-25

摘要:

网格索引构造简单,常用于数据流系统计算top k和skyline。但是,网格索引结构粗略,查询过程可能访问大量非top k结点。为了提高网格索引计算top k查询的精确度,本文提出基于数据点逆支配点集性质的网格索引方法,将查询访问集缩小到网格索引的“k最大运算区域区域kMCA”中,有效地减少了网格索引存储量和查询计算开销。同时,给出了kMCA索引结构及适应于数据流计算的kMCA维护更新算法。理论分析和实验结果均验证了上述方法的有效性。

关键词: 偏好top-k查询, 网格索引, 逆支配点集, 数据流

Abstract:

Grid index is often used in the query of top k and skyline in DSMS, but it is coarsegrained. In this paper, we propose a  Reverse Dominant Point Set   ( RDPS )  algorithm which is based on grid index, and prune a number of cells in grid index using the characteristics of RDPS to improve the precision of top k queries, thus accessing the  data set in queries is limited to  the kmax calculating region . So, it reduces the memory usage of grid index and the overhead of queries. Analytical and experimental evidences show the efficiency of the proposed approaches.

Key words: preference topk queries;grid index;reserve dominant point set;data stream