计算机科学 ›› 2010, Vol. 37 ›› Issue (11): 160-165.

• 数据库与数据挖掘 • 上一篇    下一篇

基于小波概要的区间差分skyline研究

程文聪,邹鹏,贾焰   

  1. (国防科技大学计算机学院 长沙410073);(装备指挥技术学院 北京101416)
  • 出版日期:2018-12-01 发布日期:2018-12-01
  • 基金资助:
    本文受国家863项目(2007AA01Z474,2007AA010502,2006AA01Z451)资助。

Research on Interval Differential Skyline Based on Wavelet Synopsis

CHENG Wen cong,ZOU Peng,JIA Yan   

  • Online:2018-12-01 Published:2018-12-01

摘要: 在很多应用中需要分析大量的时序数据,而相对于其它数据具有支配优势的时序数据片段往往会引起特别的关注。基于量值度量,现有的区间skyline查询可以返回给定时间区间内所有没有被其他数据支配的时序数据,这种查询有时不能满足应用的需求,且可能存在“淹没”现象。为此提出了区间差分skyline的概念,针对数据增长率属性进行分析,以解决现有区间量值skyline的不足。目前很多时序数据呈现为数据流的形式,由于资源的限制往往只会维护一个反映数据概况的概要结构,在此背景下提出了基于常用的小波概要支持不同粒度区间差分sk沙nc查询的基本算法,继而在保证准确性的基础上提出了改进后的快速算法。在真实股票价格数据集上的实验验证了所提方法的有效性。

关键词: 时序数据,区间差分skyline,小波概要

Abstract: In many applications, we need to analyze a large number of time series. Segments of time series demonstrating dominating advantages over others arc often of particular interest. Based on volume measure, the current interval skyline query returns the time series which are not dominated by any other time series in the interval. Some times this kind of query can not satisfy application rectuirements,and the "submerge" phenomenon may exist. So we proposed the concept of the interval differential skyline which focusing on the attribute of increasing rate of data to fix the shortage of the former kind of interval skyline query. Currently most of the time series are generated as data streams. Due to the limitation of the resource, people only maintain synopses which describe the main data characters. In this background we proposed the algorithm to implement the interval differential skyline query in different granularitics based on the common used wavelet synopsis and then we improved the efficiency of the naive a algorithm on the basis of keeping the accuracy of the results. Extensive experiments on the real stock price data set demonstrate the effectiveness of the proposed methods.

Key words: Time series,Interval differential skyline,Wavelet synopsis

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!