網路時代的來臨，造成了網路使用者龐大的資訊壓力，如何快速且正確的找到所需的資訊，使得資訊檢索(Information Retrieval)系統的應用成為網路世界重要的一環。隨著不同資訊型態的出現，為了適應各種型態的資訊，資訊檢索的技術也越來越多樣化，問答系統（Question Answering）是其中相當熱門的一項技術，本研究針對自動問答系統以及互動式問答系統來做探討。自動問答系統當使用者發問問題時，會嘗試從知識庫中找出正確的答案回饋，例如中文開放領域問答系統（Open-Domain Chinese Question Answering）；互動式問答系統的答案則有賴使用者問答互動產生，答案取得速度較為緩慢進而獲得更正確的答案，例如Yahoo!奇摩知識+（簡稱Yahoo!知識）。本研究將自動問答系統以及互動式問答系統優點結合，提出一個兩層式問題內容分析問答系統（Two-Level QCAS, Two-Level Question Content Answering System），。當使用者問題題目提供充足資訊可以直接擷取出答案時，則直接回覆問題答案。反之則是藉由問題內容輔助問題回答。本論文使用在Yahoo!知識所抓取的歷史人物相關問題及其回答來當作資料庫，分析已被解答的問題找出出14種問題句型以及其對應的答案句型。此外為求得更精準的資訊，加入問題內容的分析，來幫助系統找尋答案，實驗結果將Two-Level QCAS本研究將所得成果運應用在Yahoo!知識中，，替發問中問題提供解答並獲得不錯的成效，以此來驗證Two-Level 問句模型QCAS的可行性可用性。

The coming of Internet era has pressured users with enormous information. Rapidly finding necessary information has made Information Retrieval (IR) to be an important application of Internet. However, the diversity of information types evolve various IR technologies. For example, Question Answering system is one of these popular technologies. This research focuses on both Automatic Question Answering System and Interactive Question Answering System. Automatic Question Answering System, for example, Open-Domain Chinese Question Answering, retrieves answers directly from knowledge database when users request. Interactive Question Answering System, for example, YAHOO! ANSWERS, interacts with users to derive more precise answers. This research combines both advantages of Automatic Question Answering System and Interactive Question Answering System by providing Two-Level Question Content Answering System (Two-Level QCAS). When the question tiltle provides sufficient information, Two-Level QCAS replies answers directly. On the other hand, the question content is adapted by QCAS to reinforce question answering. This thesis focuses on the categories of historical characteristics in YAHOO! ANSWERS. Two-Level QCAS categorizes the corpus into 14 types of questions and the corresponding awswer patterns. In addition, question content is analyzed to help question answering The experiment shows that Two-Level QCAS applies in YAHOO! ANSWERS to obtain substantial result and verify the feasibility of Two-Level QCAS.
Keywords：Two-Level Question Content Answering System ( Two-Level QCAS)、Information Retrieval、Automatic Question Answering System

中文摘要 i
英文摘要 ii
誌謝 iv
目錄 v
表目錄 vi
圖目錄 vii
第一章 簡介 1
第二章 相關研究 5
2.1搜尋引擎 5
2.2問答系統 6
第三章 研究方法 8
3.1前置處理(Pre-processing) 10
3.1.1 語法分析(Syntactic Parsing) 11
3.2問題分析(Question Level Analysis) 12
3.3答案分析(Answer Level Analysis) 15
3.4問句的分類法 16
3.5問題類別的分類 19
3.6答案類別的分類 21
第四章 實驗 23
4.1實驗設定 23
4.2實驗結果 26
4.2.1前置實驗 26
4.2.2後續實驗 29
第五章 結論 31
5.1 結論與未來方向 31
5.2 研究限制 32
參考文獻 34
附錄 37

1. 陳鳳儀，蔡碧芳，陳克健，黃居仁，中文句結構樹資料庫的構建(Sinica Treebank)，Computational Linguistics and Chinese Language Processing Vol. 4, No 2,August, PP. 87-104
2. 張鐘尹， 漢語對話中的疑問句，國立台灣大學，語言學研究所碩士論文，民國85年。
3. Chaotao Liu, Zushu Li, “A Way of Chinese Question Analysis and Its’ Implement”, Intelligent Control and Automation, 2008. WCICA 2008. 7th World Congress on, PP. 5376-5380
4. Chen Keh-Jiann, Yu-Ming Hsieh , “Chinese Treebanks and Grammar Extraction”, Proceedings of IJCNLP-04, 2004, pp. 560-565.
5. Cheng-Wei Lee, Cheng-Wei Shih, Min-Yuh Day, Tzong-Han Tsai, Tian-Jian Jiang, Chia-Wei Wu, Cheng-Lung Sung, Yu-Ren Chen, Shih-Hung Wu, Wen-Lian Hsu, “ASQA: Academia Sinica Question Answering System for NTCIR-5 CLQA”, Proceedings of NTCIR-5 Workshop Meeting, December 6-9, 2005, Tokyo, Japan.
6. Chen Keh-Jiann, Yu-Ming Hsieh , “Chinese Treebanks and Grammar Extraction”, Proceedings of IJCNLP-04, 2004, pp. 560-565.
7. Chuan-Jie Lin, “A study on Chinese open-domain question answering systems”, Ph.D. dissertation, National Taiwan University.
8. Daniel Gildea, Daniel Jurafsky, “Automatic Labeling of Semantic Roles”, ACL 2002.
9. Dmitri Roussinov, Weiguo Fan, and José Robles-Flores, “Beyond keywords: Automated question answering on the web”, Communications of the ACM, Volume 51, Issue 9, 2008, pp. 60-65.
10. Green, B., Wolf, A., Chomsky, C., and Laughery, K., “BASEBALL: an automatic question answerer”, Readings in natural language processing, Morgan Kaufmann Publishers Inc., 1986, pp. 545-549.
11. T.-H. Tsai, S.-H. Wu, C.-H. Lee, C.-W. Shih,W.-L. Hsu. Mencius: A Chinese Named Entity Recognizer Using Maximum Entropy-based Hybrid Model. In Computational Linguistics & Chinese Language Processing 9, PP. 65-82. 2004.
12. Tianyong Hao, Dawei Hu, Liu Wenyin and Qingtian Zeng, “Semantic patterns for user-interactive question answering”, Concurrency and Computation: Practice & Experience ,Volume 20 , Issue 7 , 2008, pp. 783-799.
13. V. N. Vapnik, The Nature of Statistical Learning Theory. Springer, 1995.
14. W.-L. Hsu, Y.-S. Chen, S.-H. Event Identification Based on the Information Map – INFOMAP. In Proceedings of IEEE International Conference on Natural Language Processing and Knowledge Engineering (NLPKE), 2001.
15. Yonggang Qiu, H.P. Frei, “Concept Based Query Expansion”, SIGIR '93: Proceedings of the 16th annual international ACM SIGIR conference on Research and development in information retrieval, 1993, pp. 160-169.
16. You, Jia-Ming, Keh-Jiann Chen, “Automatic Semantic Role Assignment for a Tree Structure”, Proceedings of SIGHAN workshop, 2004.
17. ANSWERBUS, http://www.answerbus.com/index.shtml
18. ASK, http://www.ask.com/
19. CKIP AutoTag, Academia Sinica. http://ckipsvr.iis.sinica.edu.tw/
20. Google Q&A, http://blogoscoped.com/archive/2005-04-07-n20.html
21. Lucene, http://lucene.apache.org/
22. Minnesota Internet Traffic Studies , http://www.dtc.umn.edu/mints/home.php
23. Lucene, http://lucene.apache.org/
24. POWERSET, http://www.powerset.com/(已失連，抓取日期2007/7/19)
25. START, http://start.csail.mit.edu/
26. Text REtrieval Conference (TREC), http://trec.nist.gov/
27. TheEpochTimes, http://www.epochtimes.com/b5/7/3/25/n1657267.htm
28. WIKIPEDIA, http://en.wikipedia.org/wiki/Question_answering
29. WORDNET, http://wordnet.princeton.edu/
30. YAHOO！, http://developer.yahoo.com/search/boss/