Vague决策表中的含糊规则获取策略

能否有铲地处理好决策表中的含糊性在很大程度上影响到决策效果。Cau和Buehrer所提出的Vague集是对于模糊集概念的推广，最近该理论已被作为一种新型的描述含糊数据的有效工具，并在多个领域中获得了成功的应用。该文将Vague集应用于决策表中数据的描述之中，通过一个真肃属函数和一个假肃属函数来分别构造0,1]由的子区间。文中详细地讨论了造成Vague决策表中含糊性的各种因素，指出不确定性是有别于未知性的。在引入Vague区别矩阵的概念后，给出了另一种含糊规则的获取策略。通过与作者在以往文献中提出的Vague包含与相交因子的对比，从理论上分析了它们各自的适用条件。

Strategies of Ambiguous Rule Acquisition from Vague Decision Table

Fuzzy knowledge acquisition has been intensively investigated by many AI researchers. It is considered as a key problem in the fields of expert system, decision analysis, machine learning, etc. A decision table, which is a special case of information system, is a finite collection of expert data which specify what decisions should be taken when certain conditions are satisfied. Whether or not the ambiguousness can be efficiently tackled in a decision table does to a great extent affect the decision quality. However, most of the existing rule acquisition methods can not deal with the ambiguous data but only deal with the exact data. Recently, we notice that the vague set theory introduced in 1] by Gau and Buehrer has been conceived as a new efficient tool to deal with ambiguous data and it has been applied successfully in different fields. In this paper, the vague set theory is chosen to describe the ambiguous data. A vague set, as a generalization of the concept of fuzzy set, is a set of decision objects, each of which has a grade of membership whose value is a continuous subinterval of 0,1]. It is characterized by a truth membership function and a false membership function. Various factors causing the ambiguousness in a vague decision table are discussed in detail. It is emphasized that the uncertainty is different from the unknown, thus the vague similarity measures presented in 2,6], which only considered the unknown aspect of ambiguousness, do not hold any longer. Different from those measures, we propose here a more accurate similarity measure considering both aspects of ambiguousness. By introducing the concept of vague discernibility matrix, another strategy of ambiguous rule acquisition is also provided. Finally, in comparison with the strategy using the operators of vague containment and intersection introduced by the authors in 6], the idea of how to make option between these two strategies is analysed theoretically.