臺灣博碩士論文加值系統

隨著網際網路的蓬勃發展，在此新經濟時代中，客戶關係及客戶忠誠度已是企業生存的重要課題。企業若能夠於其網站上建置完善的FAQ查詢系統，不僅更可作為客服中心人員的輔助工具，以提高服務效率，更可降低客戶服務的成本；而對顧客來說，FAQ則提供其一個能夠尋找問題解答的管道，若顧客能於FAQ查詢系統上獲得滿意的解答，無形之中，也將提升其對企業的滿意度及忠誠度。
本研究建置一個在電信業領域的中文FAQ系統，其能接受使用者以口語化或近自然語言方式進行問句查詢。為了縮短系統建置的時間，我們採用文件探勘 (Text Mining) 技術來訓練分類器，在建置的流程中，先將FAQ以人工方式來建立電信業領域的階層式概念架構，接著便將所有FAQ做斷詞處理並從中萃取出重要的特徵以供學習。另外，我們利用同義詞庫，實現查詢擴張技術，以提升系統查詢的彈性及效能。藉由分類器的建立，使系統在問句查詢的過程中，能將使用者問句先行分入其所應屬的類別後，再做相似度計算，以返回答案。
在實驗評估方面，我們以相同的測試問句分別測試四個使用不同方法的系統─亦即同時使用分類器與同義詞庫的系統、同時不使用分類器與同義詞庫的系統、使用分類器但不使用同義詞庫的系統以及不使用分類器但使用同義詞庫的系統─並比較其結果。最後的實驗結果證明使用文件探勘技術確實能提升FAQ系統查詢的效能。

Following the vigorous development of the Internet, in the new era of economy, customer relationship and loyalty have become a focal point for the enterprises. If these organizations well equip a FAQ system on their web sites, it not only can assist call center employees to improve service efficiency, but also can reduce customer service cost. For customers, FAQ provides a channel to find answers to their own questions; at the same time, their level of satisfaction and loyalty to the enterprise will be increased.
We construct a Chinese FAQ system in the telecommunication domain to accept user inquiry in a colloquial or near-natural-language format. To make this possible yet efficiently, we leverage on text mining technology to train a system of classifiers. In the process, a concept hierarchy for the domain is manually made. Then, Chinese words are identified; important features from the texts such as TFIDFs are then extracted and learned. In addition, we utilize a thesauri database to enhance the efficiency and flexibility of our system. Classifiers built out of this fashion enable the system to “classify” user query to their destined answers.
We have carefully designed test queries to test in systems that use classifiers and/or thesauri database and systems that do not. Their differences are compared. Overall, experimental results demonstrate that most test queries, especially those that use text mining technology, render high accuracies with regard to finding their corresponding FAQ (answers).

中文摘要.............................................. I
英文摘要.............................................III
誌謝...................................................V
目錄..................................................VI
表目錄..............................................VIII
圖目錄................................................IX
第一章　緒論...........................................1
　第一節　研究背景與動機...............................1
　第二節　研究目的 .....................................2
　第三節　論文架構.....................................3
第二章 文獻探討........................................4
　第一節　網路上可使用之FAQ系統之簡介..................4
　　一　FAQ Finder .....................................4
二　Ask Jeeves.....................................6
　　三　寶來證券E博士..................................6
　第二節　目前FAQ系統的相關研究........................7
　第三節　中文斷詞....................................10
　第四節　特徵選取....................................12
　　一　Document Frequency Thresholding (DF)..........12
　　二　Information Gain (IG).........................12
　　三　Mutual Information (MI).......................13
　　四　χ2—Statistic (CHI)...........................14
　　五　Term Strength (TS)............................14
　第五節　文件分類技術................................16
　　一　決策樹(Decision Tree).........................16
　　二　類神經網路(Neural Network)....................17
　　三　最近鄰居(K-Nearest-Neighborhood；KNN).........19
　　四　支援向量機(Support Vector Machines；SVM)......20
　第六節　相似度計算..................................22
第三章 研究方法.......................................23
　第一節　研究方法簡介................................23
　第二節　系統建置與架構..............................24
　　一　FAQ資料之蒐集.................................24
　　二　系統架構......................................24
　第三節　系統實作....................................33
第四章 實驗與結果討論.................................38
　第一節　實驗測試資料之蒐集..........................38
　第二節　系統效能之評估準則..........................38
　第三節　實驗結果 ....................................39
　　一　問句分類之實驗................................39
　　二　問句查詢之實驗................................40
　第四節　實驗結果之討論..............................42
第五章 結論與未來展望.................................44
　第一節　研究結論 ....................................44
　第二節　研究限制....................................44
　第三節　未來研究方向................................45
參考文獻..............................................46
中文部分..............................................46
英文部分..............................................47
網站部分..............................................49
附錄A.無相對應答案的測試資料之問句分類結果............50
附錄B.問句查詢實驗之測試資料與結果....................52

中文部分
1.龐劍鋒、蔔東波、白碩，「基於向量空間模型的文本自動分類系統的研究與實現」 ，中國科學院計算技術研究所 Institute of Computing Technology ，CAS 100080，2001年。
2.林俊佑、李青松、曾廣華，「基於文件分類技術之資訊追蹤系統」，電腦與通訊，第99期，頁133。
3.劉群、李素建，「基於《知網》的辭彙語義相似度計算」，中國科學院計算技術研究所，北京大學計算語言學研究所。
4.盧增祥、李衍達，「交互支援向量機學習演算法及其應用」，清華大學學報自然科學版，1999年 第39卷 第7期 。
5.李金男，應用知網知識庫於國小課本理解之研究，成功大學資訊工程研究所碩士論文，民國89年。
6.李坤霖，網際網路FAQ檢索中意圖萃取及語意比對之研究，成功大學資訊工程研究所碩士論文，民國89年。
7.吳志鴻，應用關鍵頁搜尋及知識分類技術於Q&A系統之研究與設計，元智大學資訊管理研究所碩士論文，民國90年。
8.歐陽彥隆，電信服務業務FAQ查詢系統之研製，交通大學資訊科學研究所碩士論文，民國90年。
9.張如瑩，多語系平行關鍵頁搜尋引擎之設計與建構，元智大學資訊管理研究所碩士論文，民國90年。
10.朱毓君，以本體論強化網路FAQ系統之解答整合能力，台灣科技大學電子工程研究所碩士論文，民國90年。
11.施建新，網路資訊與學習系統之中文全文探勘工具，中央大學資訊工程研究所碩士論文，民國89年。
12.葉怡成，應用類神經網路，台北：儒林圖書有限公司，第二版，1999年。
13.中央研究院中文詞庫小組，“CKIP中文自動斷詞系統”，中央研究院，1999年。
英文部分
14.Burke, R., Hammond, K., Kulyukin, V., Lytinen, S., Tomuro, N. and Schoenberg, S., "Natural Language Processing in the FAQFinder System: Results and Prospects," In Papers from the 1997 AAAI Spring Symposium on Natural Language Processing for the World Wide Web, 1997, pp. 17-26.
15.Burke, R., Hammond, K., Kulyukin, V., Lytinen, S., Tomuro, N. and Schoenberg, S., "Question Answering from Frequently Asked Question Files: Experiences with the FAQFinder System," AI Magazine (18:2), Summer 1997, pp. 57-66.
16.Cooper, R.J., Ruger, S.M., "A Simple Question Answering System," In the Proceedings of Text Retrieval Conference (TREC-9), 2000.
17.Edwin Cooper, “Improving FAQfinder’s Performance: Setting Parameters by Genetic Programming,” AAAI Spring Symposium on MLIA Technical Papers, 1996.
18.E.H. Hovy, L. Gerber, U. Hermjakob, C.-Y. Lin, D. Ravichandran., “Toward Semantics-Based Answer Pinpointing,” DARPA Human Language Technology conference (HLT), March 2001.
19.G. Salton, “Automatic text processing : the transformation, analysis, and retrieval of information by computer,” Reading, Mass. :Addison-Wesley Publishing, 1989.
20.Hammond, K., Burke, R., Martin, C. and Lytinen, S., "FAQ Finder: A Case-Based Approach to Knowledge Navigation," In Proceedings of the 11th Conference on Artificial Intelligence for Applications. Los Angeles, CA: IEEE Computer Society Press, 1995.
21.Koza, John R., “Genetic Programming II: Automatic Discovery of Reusable Programs,” Cambridge, MA：The MIT Press, 1994.
22.Nigel Dewdney, Carol VanEss-Dykema, Richard MacMillan, “The form is the Substance：Classification of Genres in Text,” Human Language Technology and Knowledge Management.
23.Rohini Srihari and Wel Li, "A Question Answering System Supported by Information Extraction," Proceedings of ANLP, 2000.
24.Sneiders, E, “Automated FAQ Answering: Continued Experience with Shallow Language Understanding,” Question Answering Systems. Papers from the 1999 AAAI Fall Symposium. Technical Report FS-99-02, November 5-7, North Falmouth, Massachusetts, USA, AAAI Press, 1999, pp. 97-107.
25.Steven L. Lytinen, Noriko Tomuro and Tom Repede, "The Use of WordNet Sense Tagging in FAQFinder," In Proceedings of the workshop on Artificial Intelligence for Web Search at the 17th National Conference on Artificial Intelligence, Austin, TX, 2000.
26.Steven L. Lytinen, Noriko Tomuro, "TheUse of Question Types to Match Questions in FAQFinder," In Papers from the 2002 AAAI Spring Symposium on Mining Answers from Texts and Knowledge Bases, 2002, pp. 46-53.
27.Wesley T Chuang, Asok Tiyyagura, Jihoon Yang, Giovanni Giuffrida, “A Fast Algorithm for Hierarchical Text Classification,” Computer Science Department, UCLA, Los Angeles, CA 90095, USA.
28.Yang, Y., Pedersen J.P., “A Comparative Study on Feature Selection in Text Categorization,” In Proceedings of the Fourteenth International Conference on Machine Learning (ICML''97), 1997, pp. 412-420.
29.Yiming Yang and Xin Liu, “A re-examination of text categorization methods,” In Proceedings of ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR''99), 1999, pp. 42-49.
30.Zhiping Zheng, “AnswerBus Question Answering System,” In Proceeding of HLT Human Language Technology Conference (HLT2002), San Diego, CA, March 2002, pp. 24 — 27.
31.Linguistic Data Consortium (LDC), Chinese-English Translation Lexicon Version 3.0.
網站部分
32.Weka, http://www.cs.waikato.ac.nz/ml/weka/
33.Ask Jeeves, http://www.ask.com
34.AnswerBus, http://www.answerbus.com
35.寶來E博士, http://www.polaris.com.tw
36.李國維, http://www.lancom.com.tw/Express/File/express1.htm, 2001
37.台灣固網, http://www.tfn.net.tw
38.台灣大哥大, https://www.twngsm.com.tw
39.中華電信, http://www.cht.com.tw/
40.泛亞電信, http://www.tat.com.tw/
41.遠傳電訊, http://corporate.fetnet.net/index.html
42.東信電訊, http://m01ws015.mobitai.com.tw/Mobitai/main.htm
43.和信電訊, http://www.kgt.com.tw/