English  |  正體中文  |  简体中文  |  Items with full text/Total items : 51511/86795 (59%)
Visitors : 8277393      Online Users : 91
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library & TKU Library IR team.
Scope Tips:
  • please add "double quotation mark" for query phrases to get precise results
  • please goto advance search for comprehansive author search
  • Adv. Search
    HomeLoginUploadHelpAboutAdminister Goto mobile version
    Please use this identifier to cite or link to this item: http://tkuir.lib.tku.edu.tw:8080/dspace/handle/987654321/34972


    Title: 利用關聯式法則改善文件分類準確度 : 靜態與動態門檻值問題之探討
    Other Titles: Improve document classify accuracy by association rule : static threshold and dynamic threshold research
    Authors: 洪茂盛;Hung, Mao-sheng
    Contributors: 淡江大學資訊工程學系碩士在職專班
    蔣定安;Chiang, Ding-an
    Keywords: 文件分類;關聯式法則;靜態門檻值;動態門檻值;document classification;association rule;static threshold;dynamic threshold
    Date: 2009
    Issue Date: 2010-01-11 05:50:21 (UTC+8)
    Abstract: 在利用關聯式法則(Association Rule)做分類時,一般關聯式法則分類(Association Rule Classification)的信賴度門檻值設定,大多是依據經驗法則來設定單一且固定的信賴值(Confidence)為其門檻值(Threshold value),所以在設定上較為主觀。同時為了提升分類的準確度,依經驗通常選取較高的信賴度當門檻值,但門檻值若設定太高時,則容易使得部分文件因缺乏規則無法判斷其歸屬的類別,而必須利用預設規則(default rule)將這些文件分類成預設類別;若將門檻值降低則可能造成文件分類錯誤而降低分類的效能。因此本論文將針對門檻值問題做相關探討。
    本論文針對信賴度門檻值問題,分兩部分來討論,一為採取靜態門檻值或動態門檻值;另外是採單一門檻值或多重門檻值。動態門檻值的概念是在每次分類後比較準確率是否有提升,決定是否向上修定原始的門檻值;而多重門檻值的概念則是可根據不同的類別,設定不同的門檻值。
    實驗將依據不同的組合來設定門檻值,同時,希望能依據實驗的結果,找出如何能以客觀的方式來設定信賴度門檻值,並提升分類的效能。
    While using association rule for classification, the experience for association classification rules setting is following single and fixed confidence threshold value, hence is comparatively subjective. In order to increase the accuracy of classification, usually choose higher confidence in accordance with experience, but if set the confidence too high, might cause a part of documentations failed to justify the attributes by lacking rules; if set the confidence too low, it may decrease the documentation classification efficiency.
    This thesis focus on the threshold value discussion, which divides into two parts, one is static threshold value, though the training process is quicker and simpler, but during the classification procedure, the accuracy that originally already been improved could probably be influenced by follow-up lower confidence rule, namely this kind of confidence rule accuracy is low than the original threshold value setting, therefore may decrease the documentation classification efficiency, so that this thesis proposes the dynamic threshold value, to determine whether the threshold value is upward revision by after each classified whether comparative improved the accuracy or not, also propose in an objective way to set the confidence threshold value to improve the classification efficiency, this thesis proved by experiment the dynamic threshold value can obtain better classification efficiency than static threshold value.
    Appears in Collections:[資訊工程學系暨研究所] 學位論文

    Files in This Item:

    File SizeFormat
    0KbUnknown165View/Open

    All items in 機構典藏 are protected by copyright, with all rights reserved.


    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library & TKU Library IR teams. Copyright ©   - Feedback