臺灣博碩士論文加值系統

企業一直在尋找自動發佈廣告的方式來取代原先人工發佈廣告的作法來降低廣告成本，並且期望能準確的找尋潛在目標客戶發佈廣告，以提高廣告效益。然而，傳統媒體（電視、報紙及雜誌等）的廣告成本不但較昂貴，更難以鎖定特定的消費者發佈，於是企業開始尋求廣告成本較低的網際網路廣告技術。然而，過去網路廣告追蹤技術不佳，無法讓廣告商評估廣告績效，還易對使用者造成干擾。尤其是目前身受廣告商喜愛的廣告作法－E-mail行銷，所造成的垃圾資訊已經成為使用網路中的最大困擾，民眾甚至願意付費尋求可以阻絕垃圾郵件的清淨網路環境，導致網際網路廣告效益大大降低。

本論文提出一個以多代理人為基礎之網際網路廣告PULL機制，其作法是當使用者發出廣告需求的時候才會發送廣告，而且廣告的內容是經過客製化的以契合使用者需求。本系統所提出的軟體代理人不僅可攜帶這些廣告至客戶端，還可以互動的方式展現廣告內容。軟體代理人與使用者互動的歷程被蒐集、紀錄成為回饋資訊。這些回饋資訊除了能用來評估廣告的效益之外，還能以學習機制更準確的掌握使用者的喜好，促成下一次更成功的廣告。本論文並以一個e-Novel電子小說推薦系統為例，來實際展現本論文所提出的概念。

Enterprises are constantly looking for a way that helps enterprises economized by delivering ads automatically instead of formerly delivering ads by human. Furthermore, enterprises can aim accurately at target customers to delivery ads and raise benefit. However, the ad-rate of Traditional media, such as television, newspaper, and magazine, is expensive. In addition to high ad-rate, Traditional media are also difficult to aim at specific consumers and to promote advance. Thus, some enterprises interested in the Internet advertisement technology. Internet advertisement technology has cost advantage over traditional media, but the tracking technology of former Internet advertisement technology was bad. Therefore, Internet advertisement technology can not only use to evaluate benefit of publishing ads, but also bother consumers easily. Especially, E-mail, the most popular Internet advertisement technology, cause a lot of spam mail and reduce benefit.
Therefore, in our research, we propose a multi agent-based PULL mechanism for Internet advertisement systems. In this system, consumers receive ads when they ask for advertisement information and these advertisements are recommended by system to meet consumer’s need. The agents in this system are not merely bring advertisement and migrate to client site, but also interact with consumer, record the record of interacts, collection these feedback to improve recommendation mechanism and raise benefit. Also in this thesis, an e-Novel recommendation system is implemented as an example of the proposed PULL mechanism.

誌謝 i
摘 要 ii
Abstract iii
目錄 iv
圖目錄 vi
表目錄 viii
第一章 緒論 1
1.1 研究動機及目的 1
1.2 研究方法與目標 3
1.3 論文架構 4
第二章 相關研究工作 6
2.1 網際網路廣告 6
2.1.1 際網路廣告現況 6
2.1.2 網際網路使用者行為 8
2.2 PULL MODEL 與 PUSH MODEL 11
2.3 推薦的基本步驟與相關技術 15
2.4 IMMAM與SPIMMAS 27
2.4.1 IMMAM 27
2.4.2 PIMMAS 35
第三章 系統概述 41
3.1 系統設計議題 41
3.2 系統架構與重要設計議題 43
3.2.1 PULL模式網際網路廣告架構 43
3.2.2 多代理人系統 45
3.2.3 顧客特徵之擷取 51
3.2.4 廣告劇本偏好矩陣 54
3.2.5 推薦機制 57
3.2.6 學習機制 58
3.3 與前人研究之比較 61
第四章 一個實驗系統－e-Novel 推薦系統 63
4.1 實驗系統概述 63
4.2 實驗系統實作之相關技術與工具 64
4.3 小說廣告劇本格式與編寫介面 64
4.4 多代理人系統架構 69
4.4.1 讀者代理人（Reader agent） 70
4.4.2 廣告代理人（Advertising agent） 73
4.4.3 作家代理人（Writer agent） 74
4.4.4 建議代理人（Suggent agent） 74
4.4.5 學習代理人（Learning agent） 74
4.5 建構顧客特徵 75
4.5.1 資料選擇與前置處理 75
4.6 使用者劇本偏好矩陣 76
4.7 推薦與學習機制 78
4.8 實驗評估 79
4.8.1 實驗時程 79
4.8.2 推薦機制成效評估 79
第五章 結論 82
參考文獻 84

[1] M. Sloman, S. Mazumdar and E. Lupu （Eds.）, "Push vs. Pull in Web-Based Network Management," Proc. 6th IFIP/IEEE International Symposium on Integrated Network Management （IM’99）, Boston, MA, USA, May 1999, pp. 3–18, 1999
[2]Finin T, Labrou Y, Mayfield J,”KQML as an Agent communication language,” Software Agents. Cambridge, MIT Press, pp.1–22,1997.
[3]Ferreira, A.; Atkinson, J. ;”Intelligent search agents using web-driven natural-language explanatory dialogs,” Computer Vol. 38, Issue 10, pp.44 – 52, Oct. 2005
[4]P. Dasgupta, N. Narasimhan, L. E. Moser and P. M. Melliar-Smith, “A Supplier-Driven Electronic Marketplace Using mobile Agents,” Proceedings of the First International Conference on Telecommunications and Electronic Commerce, Nashville, TN, pp.42–50,November 1998
[5]Zhen-Jie Wang, Peng Ding, Huan-Ye Sheng, ” Scenario-based agent design,” Machine Learning and Cybernetics, 2003 International Conference on Vol. 1, pp.480– 484 , 2-5 Nov. 2003
[6]金凱儀, “互動式劇本為基礎之軟體代理人機制,”臺中健康暨管理學院資訊科技與管理研究所碩士論文,臺中,台灣, 2003
[7]林智偉, “SPIMMAS：以劇本與個人專屬代理人為基礎之網路行銷多代理人系統,”臺中健康暨管理學院資訊科技與管理研究所碩士論文,臺中,台灣, 2004
[8]資策會資訊市場情報中心MIC（Market Intelligence Center）URL：
http://mic.iii.org.tw/
[9]AC Nielsen, URL：
http://www.acnielsen.com.tw/
[10]M. Sloman, S. Mazumdar and E. Lupu (Eds.), “Push vs. Pull in Web-Based Network Management,”Proc. 6th IFIP/IEEE International Symposium on Integrated Network Management (IM’99), Boston, MA, USA, pp. 3–18, May 1999
使用者側寫(User profile)
[11] Rich, E. A.,” User Modeling via Stereotypes,” Cognitive Science 3[4], pp.329– 54, 1979
[12] Kun-Lung Wu, Charu C. Aggarwal and Philip S. Yu, “Personalization with Dynamic Profiler,” Proceedings of the 3rd IEEE International Workshop onAdvanced Issues of E-Commerce and Web-Based Information Systems, WECWIS, 2001.
[13] Sung Young Jung, Jeong-Hee Hong, Taek-Soo Kim, “A formal model for user preference,” Proceedings of IEEE International Conference on Data Mining, ICDM, pp. 235– 242, Dec 2002
[14] Daniel Billsus and Michael Pazzani, “Revising User Profiles：The Search for Interesting Web Sites,” Proceedings of the Third International Workshop on Multistrategy Learning, MSL , 1996
[15]黎和欣, “匿名使用者為對象之概念導向資訊推薦機制,” 中華大學資訊工程學系碩士班論文, 2001.


推薦系統
[16]J.B. Schafer, J.A. Konstan, and J. Riedl, ”E-commerce recommendation applications,” Data Mining and Knowledge Discovery, Vol.5, No.12, pp.11– 32, JAN-APR 2001
[17] Web store. Available online, URL：
http://www.amazon.com
[18]D. Goldberg, D. Nichols, B.M. Oki, and D. Terry, “Using collaborative filtering to weave an information tapestry,” Communications of the ACM, Vol.35,No.12, December 1992
[19]Paul Resnick, Neophytos Iacovou, Mitesh Suchak, Peter Bergstrom, and John Riedl, “GroupLens : An Open Architecture for Collaborative Filtering of netnews,” Proceedings of the CSCW 1994 conference,1994
[20] Mobasher, Bamshad, Dai, Honghua, Luo, Tao, and Nakagawa, Miki, “Effective Personalization Based on Association Rule Discovery from Web Usage Data,” The 3rd ACM Workshop on Web information and Data management, 2001
[21] Paul Resnick, Neophytos Iacovou, Mitesh Suchak, Peter Bergstrom, and John Riedl, “GroupLens : An Open Architecture for Collaborative Filtering of netnews,” Proceedings of the CSCW 1994 conference,1994
[22] Abraham, T. and Oliver devel, ,“Investigative Profiling with Computer Forensic Log Data and Association Rules,” Proceedings of the 2002 IEEE International Conference on Data Mining ,2002
[23] Han, J. and M. Kamber, “Data Mining: Concepts and Techniques,” Morgan Kaufmann, 2001
機器學習(Machine learning)
[24] P. Langley, H.A. Simon, “Applications of machine learning and rule induction,” Communications of the ACM 38 (11), pp.55– 64, 1995.
[25] Indranil Bose, Radha K. Mahapatra, “Business data mining-a machine learning perspective,” Information and Management, 39, pp.211– 255, 2001
[26]周寬怡,”以內容分析法獲取推薦系統中使用者profile之研究,”國立成功大學資訊管理研究所碩士論文, 2003
[27]D.E. Rumelhart, B.Widrow, M.A. Lehr, “The basic ideas in neural networks,” Communications of the ACM 37(3), pp.87–92, 1994
[28] J. Kolodner, “Case-Based Reasoning,” Morgan Kaufmann, San Mateo, CA, 1993
[29] D.E. Goldberg, “Genetic and evolutionary algorithms come of age,” Communication of the ACM 37(3), pp.113–119, 1994
[30] F. Bergadano, D. Gunetti, “Inductive Logic Programming,” The MIT Press, Cambridge, MA, 1996
[31] H. Kim and G. Koehler, “Theory and Practice of Decision Tree Induction,” International Journal Management Science on Omega, Vol.23, Issue.6, pp.587–700, Dec. 1995
[32] Tom Mitchell, McGraw Hill, ” Machine Learning”, 1997
[33]郭一聰,“應用決策樹與類神經網路於應收帳款之呆帳預警模式研究,”中原大學資訊管理學系碩士班論文, 2005
[34]陳盈妙, ”利用資料探勘於勞工退休準備金資料模式之研究,” 私立銘傳大學資訊管理研究所碩士論文, 2004
[35]顏學回;金凱儀;林志敏;,”應用劇本描述語言開發行動式代理人系統”, 第12屆行動計算研討會暨國科會行動計算計畫研究成果發表會
[36] Y.-P. Liou ,K.-Y.Chin and J. -M. Lin, "一個以多代理人為基礎之網際網路廣告PULL機制-以e-Novel推薦系統為例 ," Taiwan Conference on Software Engineering 2006,2006
[37] Ling, C.X., Tielin Chen, Qiang Yang, Jie Cheng, ”Mining optimal actions for profitable CRM”, Data Mining, 2002. ICDM 2002. Proceedings. 2002 IEEE International Conference on 9-12, pp.767 – 770, Dec. 2002
[38]Ross Quinlan,”C4.5 Programs for machine learning,” Morgan Kaufmann, 1993
[39] G.iovanni Semeraro, Marco Degemmis, Pasquale Lops, Ulrich Thiel, and Marcello L’Abbqte, “A personalized Information Search Process Based on Dialoguing Agents and User Profiling,” Proceedings of the 25th European Conference on Information Retrieval Research （ECIR ’03）, pp. 613-621, 2003
[40] R.D. Lawrence, G.S. Almasi, V. Kotlyar, M.S. Viveros and S.S. Duri, "Personalization of Supermarket Product Recommendations," Data Mining and Knowledge Discovery, 5, pp.11–32, 2001
[41]AngelCity, URL：
http://Angelcity.idv.tw
[42]JADE, URL：
http://jade.tilab.com/
[43] Microsoft Agent, URL；
http://www.microsoft.com/msagent/default.asp
[44]FIPA abstract Architecture Specification, URL
http://www.fipa.org/specs/fipa00001/SC00001L.pdf