案例式推理為近期發展的專家系統,就是將過去所發生的問題與處理方法紀錄於案例庫中,根據相似性法則由案例庫中搜尋過去近似的案例,協助新問題產生時之參考。知識工程師如何擷取專家知識是非常重要的事,然而藉由單一專家所建立的專家系統,其所得的結果往往過於主觀,但若是將多位專家的意見做有系統的彙整,又有其困難度。本研究針對多位專家的知識擷取採用三角模糊數與語意模糊這兩個主要觀念,將專家對問題點的指標權重與指標屬性間的相關程度,得以適切地定義,並以模糊綜合評價,將專家們的意見加以彙整。 本研究經驗證與分析其結果顯示為;(1)多位專家所建立之案例式推理系統比單一專家,更加客觀與完整。(2)以模糊綜合評價整合多位專家意見,可以適切地反應多位專家的知識擷取。(3)以文字比對與相似性法則對系統進行雙重檢索,可避免單以數量化為主之專家知識擷取所產生的問題。
Case-base Reasoning is a sort of Expert system developed in recently years. It stores cases in case-base database, and searches the cases in ferms of the similarity among them. It provides the aid whenever new questions or cases come up. It''s very important that how knowledge engineer used expert knowledge. There are a couple of ways used by knowledge engineer to abstract the expert knowledge. The knowledge which is abstracted by a single expert seems to be too subjective. Therefore, more than one expert be comes a beasible way, however, to do this, the integration of experts knowledge become another issues needed to be dealed with. This study is focused on two issues: one is the integration of expert knowledge by using fuzzy theory and the other is the comparision among the cases using the similarity and the linguistic variables. The results of this study reveal that (1) the system which is set up by two experts is more objective than that of only one expert; (2) the fuzzy synthesizing judgement is able to the expert knowledge appropriately; (3) the results turn out to be more usable when we use the similarity rule and the linguistic search similaritiously.