以角色為基礎於網路上進行非同步知識工程

人工智慧研究以「專家系統」的應用最為普遍,並被視為一種新的軟體技術；常見應用於醫療之專業專家系統為臨床決策支援系統(Clinical Decision Support System, CDSS)。與傳統軟體程式不同之處,建造專家系統所遭遇到最大的困難便是「專家知識的決定及編碼」,亦即知識工程(Knowledge Engineering)的過程：由知識工程師(Knowledge Engineer)引導領域專家(Domain Expert)表達及說明其專業知識(Expertise),選擇適當之推論引擎(Inference Engine)表現領域知識,將「專家知識與經驗以系統化之電腦語言表達」。其開發難度與系統本身知識密集(Knowledge-Intensive)的特性有關,包括在知識取得(Knowledge Acquisition)時,知識工程師指引專家表達及說明專業知識的過程既複雜、耗時又易於出錯,專家們若發生意見分岐或溝通困難,在選擇知識表示(Knowledge Representation)與知識整合過程時,可能產生知識庫(Knowledge Base)內的規則不完整、不一致或重覆等問題,使得知識工程過程更為困難、複雜,且整合所需時間越長。發展非同步知識工程(Asynchronous Knowledge Engineering),不僅可降低整合困難度、改善傳統知識工程師之角色功能,並不受時間、地點限制,好處多多。本研究應用已開發之web-based機率性推論診斷決策支援系統作為基本架構，提出以角色扮演模式(Role-Based Model)為基礎之非同步知識工程工具(AKET)，並透過網路環境進行非同步知識工程來建造決策支援系統，針對實作經驗進行評價。分析使用者經驗發現系統環境以使用者為中心之設計，非常簡便及具親和性，且只要經過簡單及基本的訓練即可上手；不僅解決傳統知識工程師必須與領域專家面對面同步進行知識擷取的困難度，並且透過網路可同時進行數個決策支援系統之發展，而各系統發展進度相互獨立不受影響；此外，非同步的知識工程方式更能有效解決領域專家對知識表達之分歧，排除傳統知識工程師對系統發展所造成之影響或干擾。

關鍵字

角色基礎；非同步；知識工程

並列摘要

The expert system is the most popular application of artificial intelligence (AI), and regarded as a new programming development technique. Clinical Decision Support System (CDSS) is one kind of expert system used in the medical field. In comparison with traditional software, the difficulty in the construction of such expert system lays in the knowledge acquisition, that is, the process of “Knowledge Engineering”. Knowledge engineers should guide domain experts to express their “expertise”, chose proper “Inference Engine” for knowledge representation, and yield domain knowledge and experiences in systemized computer language. The more knowledge-intensive of domain, the more difficult in developing a CDSS, especially in the stage of knowledge acquisition. During the process of knowledge engineering, it is complex, time-consumption and also easy to make mistakes. If there are conflicts or disagreements between experts, it could be very difficult in the integration of knowledge rules, data redundancy, or result in the incomplete knowledge bases. The development of asynchronous knowledge engineering tool (AKET) could reduce the difficulty of knowledge integration, which not only replace the role of knowledge engineer but also enhance the function effectively. The study purposed a role-play model of AKET upon preexisting probabilistic inference engine on the web. We constructed two CDSSs by using this model and analyzed the utilities. It proved to be user-friendly, easy-to-use, and user-oriented. The results of this study showed that AKET could resolve the problems occurred in the process of traditional knowledge engineering.

並列關鍵字

Role-based ； Asynchronous ； Knowledge Engineering

參考文獻

李友專，劉立，邱文達，〈醫學診斷決策支援系統〉Formosan J Med 1997; 1(6):722-726

薛宏昇，〈皮膚病理非感染性全身性水泡病診斷決策支援系統〉，台北醫學大學醫學資訊研究所，碩士論文，民89

張英睿，〈非感染性皮膚水泡病及自體免疫疾病之跨領域臨床診斷決策支援系統〉，台北醫學大學醫學資訊研究所，碩士論文，民90 徐恩普編著，《知識工程與專家系統》，台北:松崗，民79

電子資料行政院衛生署疾病管制局提供基層醫師做傳染性疾病鑑別診斷輔助系統http://203.65.72.83/miew/public.htm Access in 2001.11.15 Studer R, Benjamins R, Fensel D. (1998) Knowledge engineering: Principles and Methods. Available at http://www.aifb.uni-karlsruhe.de/WBS/Publ/1998/dke98.html Access in 2001.11.15

日林信成，〈基於XML之分散式模糊知識管理系統模式〉教育資料與圖書館學, 民89;37(4) 英文文獻 Newell A. The Knowledge Level. Artificial Intelligence. 1982;18:87-127.

國際替代計量

以角色為基礎於網路上進行非同步知識工程

未授權

主題瀏覽