本文將透過多功能且多層次代理人創研機制系統，來進行擴充新產業並改善學習資料庫的存取與搜尋。其多層次代理人服務包括工程代理人與商務代理人之創新研發機制，其載具目前包括CMP、動力手工具、CMP EPD 及液氣混合驅動釘槍。工程代理人包括：中介代理人輸入問題，透過工程決策代理人判斷，啟動工程創研代理人之服務及工程專利代理人搜尋專利，可透過上述代理人與TRIZ之改善與惡化參數進行創新研發新專利，並以學習方法進行資料記錄與交換;在商務代理人方面包括：將新專利透過專利分類代理人進行分類，並以專利搜尋代理人搜尋新專利相關專利，進一步至專利初步侵權相似度比對代理人比對是否侵權，最後專利鑑價代理人可輸入鑑價參數，可將工程代理人所創研出之新專利，作專利分析與價值評估，並可將分析過資料回傳至中介代理人學習資料庫中，以供使用者後續新專利研發之依據。本研究利用UML塑模技術來表達多層次代理人創研機制的知識傳遞及學習模式。
在擴充新產業代理人上，為了讓使用者在使用多層次代理人系統時，可給予使用者自行開發新產業載具，其系統提供可建立載具之本體圖之流程、工程與專利知識庫建立且由使用者依據格式輸入資料，或者外掛使用者知識庫，且額外提供使用TRIZ衝突矩陣表、TRIZ推論系統、材料與製造推論系統，來輔助研發開發新載具來達創新研發目的。本研究利用UML塑模技術進一步表達擴充新產業代理人之創新載具的流程模式。
改善記憶學習法之資料庫的存取與搜尋結構，並配合集合之屬於與交集概念來設計系統判斷邏輯，透過使用專利代理人查尋專利資料且將資料儲存至學習資料庫時，系統會自動判定關鍵字結構來自行變換順序改善存儲方式以便加快搜尋儲存資料之速度，且於搜尋資料時可指定模糊搜尋方式，搜尋之關鍵字透過模糊同義字與特徵化概念搜尋來改善搜尋結果，以達輔助了解載具之相關知識。本研究利用UML塑模技術進一步表達改善記憶學習法之資料庫存取與搜尋結構的判斷邏輯。

Through the multifunctional and multilayered agent-based innovative research and development (R&D) mechanism system, this paper expands new businesses and improves the access and search of the learning database. Its multilayered agent services include the innovative R&D mechanisms of engineering agent and commercial agent. Currently, its carriers include CMP, power-driven hand tool, CMP EPD and pneumatic-hydraulic hybrid driving nail gun.
Engineering agent includes: the import problem of Midagent; the service of EIagent and the search patent of PKagent can be started through the judgment of EDagent; innovative R&D of new patents can be conducted through the abovementioned agents as well as the improving and worsening parameters of TRIZ; and the recording and exchange of data can be done by using learning methods. As for commercial agent, it covers: the classification of new patents through the patent-classification agent; and the search of new patents and the related patents through the patent-search agent. Furthermore, patent-design-around is visited to check whether it is a tort. Finally, the patent-valuation agent can enter the valuation parameters. For the new patents innovated and studied by engineering agent, it is needed to make analysis of patents and evaluation of their prices. The analyzed information can be transmitted back to the learning database of Midagent as it can be a reference for users to make follow-up R&D of patents. The study uses the modeling technique, unified modeling language (UML) to show the knowledge transmission and learning model of the multilayered agent-based innovative R&D mechanism.
Over the expansion of new business agent, in order to let users themselves develop new business carriers during the use of the multilayered agent system, the system provides users with the procedures for building the main body drawing of carrier, and establishing the engineering and patent knowledgebase, enabling users to enter the information according to the format, or attach the users’ knowledgebase. In addition, it provides users with TRIZ conflict matrix, TRIZ inference system, as well as the material and manufacturing inference system to give assistance to the R&D of new carriers and achieve the objectives of innovative R&D. With the help of UML technique, the study further presents the procedure model for expanding the innovative carrier of the NDagent.
The access and search structures for the database that improves the memorizing and learning methods are matched with the belonging and intersection concepts of sets to design the system judgment logic. When searching the patent information and saving it in the learning database through the use of patent agent, the system can automatically judge the structure of the keywords, and then change the order and improve the form of saving, thus accelerating the searching and saving speed of information. Besides, fuzzy search can be assigned for the search of information. Through the fuzzy synonyms of the keywords to be searched and the characterized concept search, the results of search can be improved. Hence, fuzzy search helps users understand the related knowledge of the carrier. This study uses the UML technique to present the judgment logic for the access and search structures of database that improves the memorizing and learning methods.

1. Miao, Y., Li, B., Robert, G., ”An agent based application service providing model ,” Eighth International Conference on Control, Automation, Robotics and Vision, Vol. 1, pp. 120- 125(2004).
2. Wang, B. N., Gao, Y., Chen, Z. Q., Xie, J. Y., Chen, S. F. , ”A multi-agent reinforcement learning model and algorithm ,” 3rd International conference on Information Technology and Applications , pp.303-307(2005)
3. Bellifemine, F., Poggi, A., Rimassa, G., ” JADE a FIPA2000 compliant agent development environment ,” the International Conference on Autonomous, pp. 216-217(2001).
4. Wang, K. I., Abdulla, W. H., Salcic, Z., ” A Multi-Agent System for intelligent environments using JADE ,” IEE International Workshop on Intelligent Environments, pp. 86-91(2005).
5. AI-Aidaroos, H., ” Using JADE for the development of multi-agent systems ,” Measurement and Control, Vol. 38, No. 10, pp. 299-303 (2005).
6. Wang, Z. M., ” Application of an extended UML based agent-oriented in electronic commerce modeling”, Zhongshan Daxue Xuebao/Acta Scientiarum Natralium Universitatis Sunyatseni, Vol. 45, No. SUPPL., pp. 1-4(2006).
7. Park, S., Sugumaran, V., ”Designing multi-agent systems: A framework and application ,” Expert Systems with Applications, Vol. 28, No. 2, pp.259-271(2005).
8. 黃中杰、洪菁懌，JAVA與XML技術手冊，碁峰，民國九十一年。
9. 楊錦潭、蕭淳豐，「開發智慧型代理人軟體工程平台初探」，電子月刊，第六卷，第十一期，民國九十年。
10. Woodridge, M., Jennings, N. R., “ Intelligent Agents: Theory and Practice ,” Knowledge Engineering Review, Vol. 10, No. 2, p. 115(1995).
11. Genesereth, M. R., Ketchpel, S. P., “Software agents ,” Communications of the ACM, Vol. 37, No. 7, pp. 48-53(1994).
12. Shoham, Y., “Agent-oriented programming ,” Artificial Intelligence, Vol. 60, No. 1, pp. 51-92(1993).
13. Bates, J., “Role of emotion in believable agents ,” Communications of the ACM, Vol. 37, No. 7, pp. 122-125(1994).
14. Goodwin, R., “Formalizing properties of agents ,” Journal of Logic and Computation, Vol. 5, No. 6, p. 763 (1995).
15. Galliers, J. R., “A Theoretical Framework for Computer Models of Cooperative Dialogue, Acknowledging Multi-Agent Conflict , “PhD thesis, Open University, UK.(1988).
16. Rosenschein, J., Genesereth, M., “Deals among rational agents ,” Proceeding Intelligent Agents Conference Artificial Intelligence 9th, Los Angeles, pp. 91-99(1985).
17. FIPA
http://www.fipa.org/about/index.html
18. OMG
http://www.omg.org/docs/formal/00-01-02.pdf:pp.1
19. AgentBulider
http://www.agentbuilder.com/index.html
20. JATLite
http://www-cdr.stanford.edu/ProcessLink/ABE/html/jatlite.html
21. 林幸輝，「整合CMP工程創研知識及商務知識之多代理人雛型模式」，國立台灣科技大學機械工程系，民國九十四年。
22. JADEProgramming-Tutorial-for-beginners
http://jade.tilab.com/doc/JADEProgramming-Tutorial-for-beginners.pdf
23. UML Deployment Diagram
http://www.dotnetcoders.com/web/learning/uml/diagrams/deployment.aspx
24. W3C
http://www.w3.org/XML/
25. 宋瓊玲，「XML技術與圖書館應用研習班」，國立中央大學通訊第39期，第8~9頁，民國九十五年十二月
26. JDOM
http://www.jdom.org/mission/index.html
27. 陳奕彰，「整合CMP工程創研知識及商務知識之多層次代理人學習雛型模式之研究」，國立台灣科技大學機械工程系，民國九十五年。