語義排歧（Word Sense Disambiguation）一直是自然語言處理（Natural Language Processing）和語義學研究領(lǐng)域一個非常重要的課題,幾乎覆蓋了各種自然語言處理系統(tǒng),其中包括信息檢索,機器翻譯,關(guān)鍵詞的提取,語音識別,文本分類和自動文摘。語言學家和計算機科學家在探索自然語言的歧義問題上做了大量的工作,取得了很大的成績。但現(xiàn)有的對自然語言排歧的研究主要集中在語法、詞典、簡單的詞匯詞義層面上,對上下文、語義、語境、語用等知識和信息雖有涉及,但對這些信息的挖掘還相當有限。對于情態(tài)助動詞這樣語義更加模糊、對語境更為敏感的詞類的語義排歧,目前尚未發(fā)現(xiàn)。語義排歧的研究對象是語言,將語言學的研究成果用到語義排歧中,將有利于打破語義消歧的瓶頸,推動其更深入的發(fā)展。特別是在情態(tài)意義方面,從《馬氏文通》開始,語言學家對漢語情態(tài)助動詞做了深入且全面的研究。這些研究成果為構(gòu)建漢語情態(tài)動詞的語義排歧模型的設想提供了充足的理論基礎。另外,語義消歧的發(fā)展也將服務于語言學研究。情態(tài)助動詞的自動語義排歧,能夠?qū)崿F(xiàn)情態(tài)助動詞的語義的自動標注,從而為語言學家運用大規(guī)模語料庫研究情態(tài)助... 

【文章頁數(shù)】：128 頁

【學位級別】：碩士

【文章目錄】：
摘要
Abstract
Chapter 1 Introduction
    1.1 Background of this study
    1.2 Objectives of the present study
    1.3 Rationale of the study
    1.4 Outline of the present study
Chapter 2 Literature Review
    2.1 Approaches to word sense disambiguation
    2.2 Word sense disambiguation studies abroad
    2.3 Studies of WSD in China
    2.4 Studies of Chinese modal verbs
Chapter 3 Methodology and Data Collection
    3.1 Connectionism
        3.1.1 Human and Artificial Neurons
        3.1.2 Structure of Artificial Neural Network
        3.1.3 Advantages of Artificial Neural Networks
        3.1.4 The learning process
        3.1.5 Back propagation network
    3.2 Research method and data collection
Chapter 4 Sense Categorization of ‘能
    4.1 Study of sense categorization on ‘能’
    4.2 Sense categorization of ‘能’proposed by Li
        4.2.1 Epistemic modality
        4.2.2 Participant-internal modality
        4.2.3 Participant-external modality
        4.2.4 Deontic modality
Chapter 5 The Construction of the Neural Network Model for Word Sense Disambiguation of ‘能’
    5.1 WSD Model
    5.2 Construct two sense-tagged corpora of ‘能’
    5.3 Feature extraction
    5.4 Transfer linguistic features into vectors
        5.4.1 Define the input vectors
        5.4.2 Define the output vectors
    5.5 Determine the number of nodes in hidden layer
    5.6 Model constructing in Matlab
        5.6.1 About Matlab
        5.6.2 Construct neural network model in Matlab
    5.7 Summary
Chapter 6 Further Study
    6.1 Distribution of four senses of ‘能’in corpora
    6.2 Co-occurrence of ‘能’with linguistic features
        6.2.1 Co-occurrence of ‘能’with syntactic features
        6.2.2 Co-occurrence of ‘能’with semantic features
    6.3 The contribution of semantic features and syntactic features to word of ‘能'
    6.4 Summary
Chapter 7 Conclusion
References
Appendix I
Appendix Ⅱ
Appendix Ⅲ
Appendix Ⅳ
Acknowledgements
作者簡介


【參考文獻】：
期刊論文
[1]基于可拓學理論的漢語詞義消歧[J]. 盧志茂,劉挺,李生.  哈爾濱工業(yè)大學學報. 2006(12)
[2]基于多分類器決策的詞義消歧方法[J]. 全昌勤,何婷婷,姬東鴻,余紹文.  計算機研究與發(fā)展. 2006(05)
[3]利用BP神經(jīng)網(wǎng)絡的中文詞義消歧模型[J]. 何婷婷,謝芳.  華中師范大學學報(自然科學版). 2005(04)
[4]基于MDL聚類的無導詞義消歧[J]. 陳浩,何婷婷,姬東鴻.  小型微型計算機系統(tǒng). 2005(10)
[5]情態(tài)與漢語情態(tài)動詞[J]. 朱冠明.  山東外語教學. 2005(02)
[6]情態(tài)動詞“能”在交際過程中的義項呈現(xiàn)[J]. 王偉.  中國語文. 2000(03)
[7]基于神經(jīng)網(wǎng)絡的漢語口語多義選擇[J]. 王海峰,高文,李生.  軟件學報. 1999(12)
[8]助動詞“想”和“要”的區(qū)別[J]. 張維耿.  語言教學與研究. 1982(01)



本文編號：3652221