本論文藉由對一般使用者問題進行觀察與分析，提出淺層語意問題分析模型與淺層語意答案推論模型來處理一般使用者發問的醫學相關問題，我們考慮到一般使用者在非醫學專業背景的情況下，以使用者問題中的關鍵詞彙，分成Entity與Feature兩種淺層語意角色進行問題分析，更利用搜尋引擎Google來替我們找到網路上能回答使用者問題的相關文件，接著透過我們提出的淺層語意答案推論模型，藉由我們利用美國國家醫學圖書館(NLM)提供的分類架構來建立我們的模型，並以國家網路醫學圖書館(Kingnet)內的醫學相關文件進行概念詞的蒐集及對應，最後利用訓練完成的淺層語意答案推論模型來由相關文件中找出相對應於使用者問題的詞組層級答案、句子層級答案及段落層級答案提供給使用者做參考。目前在中文的問答研究方面就我們所知並沒有針對一般使用者設計的醫學專業問答研究，因此我們在評估方面，並沒有一個比較好的參考對象，我們只能就我們系統的效能來進行評估，我們針對詞組層級答案、句子層級答案及段落層級答案分別進行評估，並分別得到0.553、0.5771及0.5886的MRR評估結果，雖然我們找不到一個較接近的系統進行比較，不過就目前在中文醫學專業資源仍不充足的情況下，我們試著利用本論文提出的兩個模型，搭配上現有的醫學資源，如MMODE、中英雙語MeSH概念詞，以及由美國國家醫學圖書館提供的NLM Classification，建構一個中文自然語言醫學問答系統，來解決一般中文使用者提出的醫學相關問題。

This paper proposes Shallow Semantic Question Analysis Model and Shallow Semantic Answer Inference Model to address some general medical questions according to the observation and analysis of user questions. Considering the condition that general users are not in the background of medical expertise, we analyze questions by identifying both shallow semantic Entity and Feature according to the keywords of users’ questions. Moreover, we use Google to find relevant documents which can answer user questions. Then, we propose Shallow Semantic Answer Inference Model using the classification architecture provided by National Library of Medicine (NLM) to construct our models and match the conceptual keywords with medical related documents in Kingnet. Finally, we utilize the trained Shallow Semantic Answer Inference Model and relevant documents to find the corresponding phrase-level, sentence-level, and paragraph-level answers for users.
In Chinese QA research until now, there is no automatic medical QA system developed for general users. Therefore, there is no better benchmark for evaluation. We can only evaluate the performance of our system. We utilize MRR as our evaluation metric and evaluate the accuracy of phrase-level answers, sentence-level answers and paragraph-level answers, and obtain a result of 0.553, 0.5771, and 0.5886, respectively. Although there is no similar system for our comparison, however, without sufficient Chinese medical resources, we try to utilize two proposed models and existing medical resources, such as MMODE, Chinese-English bilingual MeSH concept terms, and NLM Classification to build a Chinese medical QA system which can deal with users’ natural language questions.

M. W. Bilotti, P. Ogilvie, J. Callan, E. Nyberg 2007. Structured Retrieval for Question Answering. In Proc. of SIGIR 2007: 351-358.

E. Brill, S. Dumais, Michele Banko 2002. An Analysis of the AskMSR Question-Answering System. Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMLP), Philadelphia, July 2002, pages 257-264.

Y. Chen, M. Zhou, S. Wang 2006. Reranking Answers for Definitional QA Using Language Modeling. Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the ACL, pages 1081–1088.

C. L. A. Clarke, G. V. Cormack, T. R. Lynam 2001. Exploiting Redundancy in Question Answering. In Proc. of SIGIR 2001: 358-365.

J. Chu-Carroll, K. Czuba, J. Prager, A. Ittycheriah 2003. In Question Answering, Two Heads Are Better Than One. In Proc. of HLT-NAACL: 24-31.

H. Cui, R. Sun, K. Li, M. Y. Kan, T. S. Chua 2005. Question Answering Passage Retrieval Using Dependency Relations. In Proc. of SIGIR 2005: 400-407.

D. Demner-Fushman, J. Lin 2006. Answer Extraction, Semantic Clustering, and Extractive Summarization for Clinical Question Answering. Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the ACL, pages 841–848.

S. Dumais, M. Banko, E. Brill, J. Lin, Andrew Ng 2002. Web Question Answering: Is More Always Better. In Proc. of SIGIR 2002: 291-298.

H. Fang, C. Zhai 2006. Semantic Term Matching in Axiomatic Approaches to Information Retrieval. In Proc. of SIGIR 2006: 115-122.

J. Ko, T. Mitamura, E. Nyberg 2007. Language-independent Probabilistic Answer Ranking for Question Answering. Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics, pages 784–791.

J. Ko, L. Si, E. Nyberg 2007 A Probabilistic Graphical Model for Joint Answer Ranking in Question Answering. In Proc. of SIGIR 2007: 343-350.

K. W. Kor, T. S. Chua 2007. Interesting Nuggets and Their Impact on Definitional Question Answering. In Proc. of SIGIR 2007: 335-342.

A. Korhonen, Y. Krymolowski, N. Collier 2006. Automatic Classification of Verbs in Biomedical Texts. Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the ACL, pages 345–352.

M. Lee, J. Cimino, H. R. Zhu, C. Sable, V. Shanker, J. Ely, H. Yu 2006. Beyond Information Retrieval—Medical Question Answering. AMIA 2006 Symposium Proceedings, pages 469-473.

J. Lin 2006. The Role of Information Retrieval in Answering Complex Questions. Proceedings of COLING/ACL Main Conference Poster Sessions, pages 523-530.

J. Lin and D. Demner-Fushman 2006. The Role of Knowledge in Conceptual Retrieval: A Study in the Domain of Clinical Medicine. In Proc. of SIGIR 2006: 99-106.
J. Lin and B. Katz 2003. Question Answering from the Web using Knowledge Annotation and Knowledge Mining Techniques. In Proc. of CIKM 2003: 116-122.

A. Moschitti, S. Quarteroni, R. Basili, S. Manandhar 2007. Exploiting Syntactic and Shallow Semantic Kernels for Question/Answer Classification. Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics, pages 776–783.

Y. Niu, X. Zhu, G. Hirst 2006. Using Outcome Polarity in Sentence Extraction for Medical Question-Answering. AMIA 2006 Symposium Proceedings, pages 599-603.

A. Novischi, D. Moldovan 2006. Question Answering with Lexical Chains Propagating Verb Arguments. Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the ACL, pages 897–904.

B. Ofoghi, J. Yearwood, R. Ghosh 2006. A Semantic Approach to Boost Passage Retrieval Effectiveness for Question Answering. Proceedings of the 29th Australasian Computer Science Conference, Volume 48.

M. Pasca 2007. Lightweight Web-Based Fact Repositories for Textual Question Answering. In Proc. of CIKM 2007: 87-95.

J. Prager, P. Duboue, J. Chu-Carroll 2006. Improving QA Accuracy by Question Inversion. Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the ACL, pages 1073–1080.

S. L. Price, L. M. Delcambre, M. L. Nielsen 2006. Using Semantic Components to Express Clinical Questions Against Document Collections, In Proc. of HIKM 2006: 9-16.

G. Ramakrishnan, S. Chakrabarti, D. Paranjpe, P. Bhattacharyya 2004. Is Question Answering an Acquired Skill?, In Proc. of WWW 2004: 111-120.

S. Tellex, B. Katz, J. Lin, A. Fernandes, and G. Marton 2003. Quantative Evaluation of Passage Retrieval Algorithms for Question Answering, In Proc. of SIGIR 2003: 41-47.

M. Wu and T. Strzalkowski 2006. Utilizing Co-Occurrence of Answers in Question Answering. Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the ACL, pages 1169–1176.

H. Yang, T. S. Chua, S. Wang, C. K. Koh 2003. Structured Use of External Knowledge for Event-based Open Domain Question Answering, In Proc. of SIGIR 2003: 33-40.

H. Yu 2006. Towards Answering Biological Questions with Experimental Evidence: Automatically Identifying Text that Summarize Image Content in Full-Text Articles. AMIA 2006 Symposium Proceedings, pages 834-838.

W. Zhou, C. Yu, N. Smalheiser, V. Torvik, J. Hong 2007. Knowledge-intensive Conceptual Retrieval and Passage Extraction of Biomedical Literature. In Proc. of SIGIR 2007: 655-662.