知识粗糙性的粒度原理及其约简  

Principle of granularity of knowledge roughness and reduct computing

作　　者：耿志强[1] 朱群雄[1] 李芳[1] 

机构地区：[1]北京化工大学信息科学与技术学院,北京100029

出　　处：《系统工程与电子技术》2004年第8期1112-1116,共5页Systems Engineering and Electronics

基　　金：教育部科学技术研究重点项目基金资助课题(01024)

摘　　要：粗糙集理论是一种新的软计算方法,已成为知识发现和诊断决策领域的一个研究热点。经典的粗糙集理论提出知识是有粒度的并定义了知识粗糙度的概念,但它不能完全区分不同信息粒度所表示的信息量。从信息论的角度定义了信息粒度的概念,重点研究了知识粗糙性的粒度原理,定义了粒度函数和粒度熵的概念,提出了信息粒度的量化计算方法,解决了知识粗糙度在表达信息时的不足。根据知识粗糙性和信息粒度本质上的一致性,提出了一种基于粒度熵的属性约简算法,该算法可以从各约简集中选择最优属性约简,避免了选择约简集的盲目性。实例研究证明提出的粒度计算方法是可靠有效的,为进一步研究知识的粒度计算提供了可行的方法。Rough set theory is a relatively new soft computing technique and has become a topic of general interest in the field of knowledge discovery and decision-making. The classical rough set theory thinks knowledge is granular and defines the concept of knowledge roughness, but it can not completely distinguish the information from the different knowledge granularity. From the information theory, the concept of information granularity is defined. Granularity principle of knowledge roughness is studied stressfully. Granularity function and granularity entropy are defined and the quantitative computing method based on information granularity is proposed, so the shortage of knowledge roughness is solved. According to the consistency between knowledge roughness and informaton granularity, a new attribute reduction algorithm based on granularity entropy is proposed. The optimal reduction set can be selected from all reduction sets and the blindness of selecting reduction set is avoided. The validity of proposed granularity computing algorithm is proved by the application of practical database. Moreover, it puts forward new methods for further researching granularity computing of knowledge.

关 键 词：软计算 信息论 粗糙集 知识发现 信息粒度 粒度熵 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]