skip to main content
資源種類 顯示結果: 顯示結果: 查詢種類 索引

一個使用漢明基礎編碼法來有效提供在資料方體上的部分和查詢之設計與分析

鄭兆晏 李建億; 資訊教育研究所 2001

線上取得

  • 題名:
    一個使用漢明基礎編碼法來有效提供在資料方體上的部分和查詢之設計與分析
  • 著者: 鄭兆晏
  • 李建億; 資訊教育研究所
  • 主題: 資料方體; 漢明碼; 部分和查詢; data cube; hamming code; Parital-Sum Queries
  • 描述: 摘 要 由於線上分析處理(OLAP)的廣泛應用,使用者可以從資料倉儲(data warehouse)中分析集結性的資料。一個典型的線上分析處理架構是由許多方格所組成的資料方體(data cube)。為了符合資料方體的即時性需求,如何加速總集性資料查詢是一個值得研究的議題。總集性資料查詢主要可分為兩種:一是連續性的資料和查詢,稱為範圍和查詢(range-sum query)。一是不連續性的資料和查詢,稱為部分和查詢(partial-sum query)。本論文主要針對部分和查詢作研究。已知的部分和查詢方法是結合覆蓋碼的特性,以查表的方式解部分和問題。其主要的缺點是浪費很多空間存放索引查詢表格。所以,在本論文中,我們將提出HBC(Hamming-based Code)法以有效解決部分和問題。其主要的構想是藉由漢明編碼(hamming code)的方式來建立種子表格,如此,不但查詢時可以利用其偵錯的功能快速找出種子和錯誤項,而且可以節省下存放索引查詢表格的空間。此外,我們將擴展HBC法以突破漢明碼只能偵錯一個項目的限制,且使用者可依個別需求建立不同的種子表格。對於動態和維護問題,種子表格不需重建只要更動部分的種子項即可。再則,本論文所提出的方法可以擴展以解決多維度(multi-dimensional)部分和問題。最後,我們結合機率問題提出改進方法使查詢效率更佳,並同時以數學和實驗方式進行分析比較,結果證實本方法具有較佳的存取效能。
    Abstract For the general application of OLAP, users can analyze aggregate data from the data warehouse. A typical structure for OLAP is a data cube, which is composed of cells. Because of the real-time request in a data cube, how to speed up aggregate values of some specified cells demanded from a query is an important research topic. The types of an aggregate data query can be divided into two ways. One is the continuous data query, called a range-sum query. The other one is the discrete data query, called a parital-sum query. We focus our research in the latter one. The proposed methods to solve a parital-sum query are using a covering code approach and creating the index look-up table. However, their major drawback is wasting too much memory space for the index look-up table. In this thesis, we will propose an efficient method, called HBC(Hamming-based Code)method, for supporting a partial-sum query. The main idea is to establish a seed table by applying hamming-based codes. Via the seed table that, we can use its error-detection function to find the seed and error bit quickly in a query process, and save much memory space. Moreover, we will also extend the HBC method to handle mutiple-error problem, such that the users are able to establish a distinct seed table according to their request. For dynamic data cubes, the seed table only needs to change some part of seeds without reconstruction. Furthermore, we can expand the HBC method for the multidimensional partial sum problem. Finally, we will consider the probability of query on each cell to improve access performance. Our method will be evaluated using analytical cost model, and the analytical results indicate that it has a better performance than the others.
    碩士
  • 建立日期: 2001
  • 格式: 121 bytes
    text/html
  • 語言: 中文
  • 識別號: http://nutnr.lib.nutn.edu.tw/handle/987654321/4860
  • 資源來源: NUTN IR

正在檢索遠程資料庫,請稍等