多准则分级决策的扩展粗糙集方法

解决分级决策问题的方法主要涉及统计学、人工智能和运筹学等．粗糙集理论被证明是进行多属性决策分析的有利工具．但是，基于不可分辨关系或相似关系的传统粗糙集方法不适于解决带有准则的决策分析问题．因此，Greco等提出了一种扩展的粗糙集方法分析分级决策数据，该方法利用支配关系代替不可分辨关系对决策类进行粗糙近似．在此基础上，为了从决策数据中构造偏好模型，通过构造支配矩阵和支配函数计算最小决策规则．为了消除规则集中的冗余性，提出了规则化简的方法．此外，对基于规则的分级决策策略进行了研究.

Multi-criteria sorting decision based on extended rough sets theory

Methodologies for sorting problems have been developed from a variety of research disciplines, including statistics, artificial intelligence and operations research. Rough set theory has proved to be a useful tool for analysis of a large class of multi_attribute decision problems. But, the theory based on indiscernibility relation or similarity relation can not handle the decision problems with criterion. Greco et al have proposed a kind of extended rough set approach, in which the rough approximations of decision classes involve using dominance relation instead of using indiscernibility relation in the basic rough sets approach in the analysis of sorting examples. In order to construct preference models from decision examples, the dominance function is constructed to compute the minimal decision rules. Then, the methodology of rule simplification is presented to eliminate the redundancy in the rule set. In addition, the strategies of sorting decision based on the derived rules are put forward.