對於一個企業而言，供應商的選擇是相當重要且複雜的決策問題，採購決
策者通常須要同時考慮不同評估準則後，才能作出正確的抉擇。在複雜產業相
互競爭環境中，這些評估準則存在具有相互衝突、不確定性或不可量化之因
素，由於資訊的不精確及定量與定性的特性，往往造成決策者無法建立客觀與
具體量化的評估架構。本文利用證據推論的方法，發展出結合一套定性與定量
之證據推論評估模式，此一評估模式利用理想解類似度偏好順序技術(TOPSIS)
作為將不確定性定量指標轉換為基本機率指派函數，利用丹普斯特-雪佛層級
分析法作為將定性指標轉換為基本機率指派函數，最後再利用丹普斯特-雪佛
理論合併所有相關證據，以排序優先選擇方案，此一柔性計算分析方法可協助
企業進行供應商評選決策制定時參考應用。本文除詳述所提出的四個分析步驟
外，並舉出一個案例說明如何使用此一評估模式。

Selecting superior vendors is a complex decision task. It requires a large number of criteria to be measured and evaluated simultaneously. However, many of these criteria are uncertain and often conflict with each other. Furthermore, since the criteria under evaluation are expressed either in quantitative or qualitative terms, decision-makers have to consider their
judgments based on the both quantitative data and experiential subjective
assessments. In this paper, the evidential reasoning (ER) approach (which
is capable of processing both quantitative and qualitative measures) is applied to solve the vendor selection problem (VSP). In this study, the process of building a multiple criteria decision model using the ER approach is presented, in which both quantitative and qualitative information is ex-pressed in a unified framework by incorporation with TOPSIS and Dempster-Shafer-analytic hierarchy process (DS-AHP). A four-step procedure is proposed to evaluate the VSP and the practical usage is illustrated by a given example.