在高度資訊化社會中，以自然語言輸入方式之對話系統，是未來最理想的人機互動方式之一。在對話系統中，錯誤的語意理解常使得人機對話無法順利進行，所以如何讓電腦了解語者意圖成為一個值得研究的主題。
　　本論文主要是提出語意相依關係應用於語意理解，目的是將深層的語意結構解析出來幫助語意分析，其精神為除了考慮語句中詞的表徵意義外，還進一步將詞與詞之間所隱含之語意關係表達出來。而在對話過程中使用語意相依圖取代語意框架，減少因誤判意圖而造成對話狀態資訊之遺失。另外還加入對話歷史關連，擷取對話之語意脈絡，使得對話過程更為流暢。語意相依關係之擷取主要是根據句子的詞組結構和語意概念資訊，所以在發展系統時我們採用中研院TreeBank和知網(HowNet)作為系統知識庫。
　　為了進行方法之評估，首先建立一個醫療服務對話系統作為實際應用測試平台，此系統主要是提供掛號和科別查詢的服務，在測試系統時，分別就每項意圖做測試，發現整體意圖偵測正確率為95.6%、對話完成率為85.24%，每筆對話的平均長度為8.3回合，相對於應用貝氏分類器，部分樣本樹搭配語意框架的方法分別提升了14.9%、12.47%。由實驗可知論文所提之方法在實際應用上其效能都能有明顯的提升。

　　In the high information-intensive society, one of the most ideal man-machine interactive communications is the dialogue system using natural language in the near future. The misunderstanding in the semantic interpreter usually result in the un-complete dialog in the traditional dialog management, especial in the speech act or intention identification. The understanding of the utterance of the user will become the most interesting research issue.
This thesis mainly proposes a novel understanding approach called by Semantic Dependency Analysis (SDA), which purpose is to find the implicit semantic dependence between the concepts. Besides definitions of the semantic concepts, the dependence structures between concepts in the utterance are also took into consideration. Instead of semantic frame/slot, SDA can keep the more information when the system can not clearly identify speech act or intention.
　　This thesis also uses dialogue history to help understanding the utterances. The Semantic Dependency Relations are built according to the structure of sentence and the conceptual meaning of the words. When developing of the system，we use Sinica TreeBank and HowNet as the system knowledge.
　　In order to evaluation the method we proposed, the medical service dialog system is developed. The accuracy rate of speech act detection is 95.6%, the task-completion rate is 85.24%, and the average number of turns of each conversation is 8.3. Compared with the Bayesian Classifier and Partial-Pattern Tree based approaches, we have 14.9% and 12.47% improvement in accuracy rate of Speech Act respectively. The result showed that the performance of the proposed method is obviously improved, namely the SDA approach outperforms the traditional approaches in the semantic understanding of spoken dialog system.

[1] Michael F. McTEAR, “Spoken Dialogue Technology: Enabling the Conversational User Interface,” ACM Computer Surveys, Vol 34, No. 1, March 2002, pp.90-169.
[2] J. F. Allen, D. K. Byron, Dzikovska M. Ferguson, L. Galescu, and A. Stent, (2001) “Towards Conversational Human-Computer Interaction,” AI Magazine.
[3] Xuedong Huang, Alex Acero, Hsiao-Wuen Hon, “Spoken Language Processing”, Prentice-Hall,Inc. 2001
[4] Ferguson, George M., Allen, James F., Miller, Brad W., Ringger, Eric K., "The Design and Implementation of the TRAINS-96 System: A Prototype Mixed-Initiative Planning Assistant", TN96-5, Computer Science Dept., U. Rochester, October 1996.
[5] Masahito Kawamori, Akira Sugiyama, and Takeshi Kawabata: Handling Rich Turn-Taking in Spoken Dialogue Systems,Eurospeech99, Proc. of Eurospeech99, pp.1167-1170, 1999-09
[6] AT&T(2002)How May I Help You ?
http://www.research.att.com/~algot/hmihy/
[7] S. Bennacef and L.Lamel et al., "Dialogue in the RAILTEL Telephone-Based System," ICSLP'96 Vol. 1. pp. 550-553
[8] L. Lamel, S. Rosset, J. Gauvain, S. Bennacef, M. Garnier-Rizet , and B. Prouts, (2000) "The LIMSI Arise System," Speech Communication, 31(4), pp. 339-353.
[9] Chun-Jen Lee, Eng-Fong Huang, and Jung-Kuei Chen, "A Multi-keyword Spotter for the Application of the TL Phone Directory Assistant Service," Proceedings of 1997 Workshop on Distributed System Technologies & Applications, pp. 197-202
[10] Tung-Hui Chiang, Chung-Ming Peng, Yi-Chung Lin,Huei Ming Wang and Shih Chieh Chien, “The Design of a Mandarin Chinese Spoken Dialogue System,” in Proceedings of COTEC’98, Taipei 1998, pp. E2-5.1~E2-5.7
[11] 陳銘軍, 葉瑞峰, 吳宗憲, “以知識概念模型為基礎之多主題對話管理系統,” in Proceedings of ROCLING XV, Hsinchu, Taiwan, 2003.
[12] J. Allen, (1994) “Natural Language Understanding”, The Benjamin/Cummings Publishing Company
[13] J. F. Sowa, Knowledge Representation: Logical, Philosophical and Computational Foundations, 1999, Brooks Cole Publishing Co.
[14] Wang,Y.-Y. and A Acero. Combination of CFG and N-gram Modeling in Semantic Grammar Learning.Eurospeech.2003.Geneva
[15] Wang,Y.-Y. Robust Spoken Language Understanding in Mipad. Eurospeech.2001.Aalborg,Denmark.
[16] 陳克健、黃居仁《中文句結構樹資料庫的構建(Sinica Treebank)》
[17] 董振東, 董強, “知網”, http://www.keenage.com/
[18] Meng,H. and Siu, K. C.(2001) “Semiautomatic Acquisition of Semantic Structures for Understanding Domain-Specific Natural Language Queris,”IEEE Transaction Knowledge and Data Engineering , pp.172-181
[19] 中研院詞庫小組技術報告93-05中文詞類分析
[20] Detlet Prescher “A Tutorial on the Expectation-Maximization Algorithm Including Maximun-Likelihood Estimation and EM Training of Probabilistic Context-Free Grammars” University of Ansterdam0. 2003.
[21] Traum, D.R. "Speech Act for Dialogue Agents"pp.169-201,Kluwer Academic Publishers,1999March 2002, pp.90-169.
[22] M. A. Walker, D. Litman, C. Kamm, and A. Abella, 1997. PARADISE: a general framework for evaluating spoken dialogue agents. In Proceedings of the 35th Annual General Meeting of the Association for Computational Linguistics, ACL/EACL (Madrid, Spain). ACL, 271–280.
[23] Chung-Hsien Wu, Yeou-Jiunn Chen, and Cher-Yao Yang, ”Error Recovery and Sentence Verification Using Statistical Partial Pattern Tree for Conversational Speech,” in Proceedings of ICSLP2000, Beijing, China, 2000.
[24] Duda, R.O. and P.E. Hart. Pattern Classification and Scene Analysis, 1973, New York, Wiley.