广义不完备序值信息系统中的知识约简  

Dominance-based Rough Set Approach and Knowledge Reductions in Generalized Incomplete Ordered System

作　　者：莫京兰[1] 朱广生[2] 吕跃进[3] 

机构地区：[1]广西科技大学鹿山学院公共教学部,广西柳州545616 [2]广西科技大学校团委,广西柳州545006 [3]广西大学数学与计算科学学院,南宁530004

出　　处：《小型微型计算机系统》2015年第12期2735-2739,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(71361002)资助;广西自然科学基金项目(2013GXNSFAA019016)资助;广西高等学校立项科研项目(2013LX095;YB2014501)资助

摘　　要：以同时具有丢失型和遗漏型未知属性值的广义不完备序值信息系统为研究对象,提出确定特征优势关系的概念,并基于此关系讨论了该系统的知识约简问题.对比分析证明,确定特征优势关系在一定程度上克服了特征优势关系的局限性.在广义不完备序值决策系统中引入近似分布约简,提出新的近似分布约简概念,结果表明基于确定特征优势关系粗糙集模型能有效地处理具有两种未知属性值的广义不完备序值系统,为从复杂的不完备系统中知识获取提供了新的理论基础与技术手段.In this paper, definitive characteristic dominance relation was defined for the classification of universe in the generalized in- complete ordered information systems in which lost and ＂do not care＂ unknown attribute values coexist. Based on the definitive char- acteristic dominance relation ,knowledge reductions ware discussed in this information systems. Through comparison and analysis, de- finitive characteristic dominance relation overcome limitations of characteristic advantage in a certain extent. Furthermore, the approxi- mate distribution reduct is introduced into generalized incomplete ordered decision systems and then new approximate distribution re- ducts ware proposed. This research was used to show that rough set model based on the characteristic dominance relation was effective in dealing with the generalized incomplete ordered system that contained two different types of unknown values. Then,it provided new theoretical basis and the technical means for knowledge acquisition in complex incomplete information system.

关 键 词：不完备信息系统 广义不完备序值信息系统 优势关系 确定特征优势关系 知识约简 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]