企業網站評估之模糊積分階層式多屬性決策

Abstract

在現實的世界裡，大部分決策之準則彼此間都會有關係，因此使用傳統的加法測度較不合適。而模糊測度(fuzzy measure)則將假設放寬，僅要求滿足單調性，且使用不需要假設可加性和獨立性的模糊積分(fuzzy integral)來處理人類主觀評估程序是較為適當的。本研究修改了一個演算法，只需輸入模糊密度(fuzzy densities)就可獲得l-值，且利用以l-模糊測度為基礎的模糊積分去獲得整體評價值，最後並以一個企業內部網站評鑑之實例來對階層式多屬性決策做個說明，而結論也得到在處理人類主觀評估或準則間不獨立時採用模糊積分是較傳統的多屬性評估來得適合。

In the real world, most of criteria have inter-dependent characteristics and cannot be evaluated by the conventional additive measure. However, it would be more appropriate to apply a fuzzy integral model, in which it is not necessary to assume additivity and independency, to approximate the human subjective evaluation process. This research modifies an effective algorithm to determine the l-value by the input data of fuzzy densities and by fuzzy integral based on l-fuzzy measure to determine the overall evaluation. Finally, this research gives an example of enterprise intranet web sites evaluation with illustrations of hierarchical structure of l-fuzzy measure for Choquet integral model, and the results show fuzzy integral is more suitable than a traditional multi-criteria evaluation method for human subjective evaluation or criteria are not mutually independent.