Abstract: In view of the decision making problems that the Decision Maker（DM） provide their preference information by using linguistic labels, the definitions of order consistency and additive consistency of Linguistic Preference Relations（LPRs） are introduced. Then, the determination methods of additive consistent LPRs are studied, and the additive consistent induced?LPR is investigated. The concepts of consistency index and acceptable consistency of LPRs are also presented. In addition, based on the improving the consistency for LPRs, a decision making algorithm is developed to adjust unacceptably consistent LPR into an acceptably one, and the convergence of the proposed algorithm is studied. Finally, a practical application to selection of database system is demonstrated to show the rationality and effectiveness of the proposed algorithm.

Key words: linguistic preference relation, order consistency, additive consistency, consistency index, decision making algorithm, database

摘要： 针对决策者提供的偏好信息为语言标量的决策问题，首先引入了语言偏好关系的有序一致性和加性一致性的定义，研究了语言偏好关系加性一致性的判定方法，构建了满足加性一致性的诱导语言偏好关系，提出一致性指数和满意一致性的概念；然后建立了基于语言偏好关系一致性改进的决策算法，并证明了算法的收敛性，同时通过该算法改进后的语言偏好关系满足满意一致性条件。最后通过数据库系统的选择实例说明提出的决策算法是合理的和有效的。

关键词: 语言偏好关系, 有序一致性, 加性一致性, 一致性指数, 决策算法, 数据库