基于粗规则对象空间信息表的最小规则集生成

提出了粗规则对象空间信息表的概念、相关概念和性质，并给出实现粗规则最小化算法。基于此，编写了相应的计算机程序，并对一个地理信息系统实例规则进行了研究。实例表明运行可靠、结果正确，可应用于智能决策信息系统的知识库优化。

Generation of Minimum Rules from Rough Rule Sets Based on Object-spatialInformation Table

The theory of Rough sets has been widely used in such fields as data mining and knowledge discovery.We need to explore a great amount of potential information contained in databases.Among those databases,some can be regarded as rough knowledge base,which needs to be further processed into least rule set(LRS).In this paper,some definitions of Rough sets were outlined.Based upon the theory of Rough sets,this article gave the concept of object spatial information table(OSIT) of rough rules,introduced some concepts concerning object spatial information table,gave a practical algorithm to realize the reduction of redundant rules in rough rule set,and suggested a computer program using C language.An example in the field of GIS was studied.Object is one record of the original rough set and object spatial information table is a set of records which has a similar table form as the original set.OSIT gives the complete information for the object.The content of OSIT is determined by both the content of the object itself and the original rough set.All objects in the rough set have the same table form derived from the original rough set but they are different in content.Because of this,computer program can be easily processed to reduce the redundancy of a given rough rule set.This article detailed in the processes of the generating LRS for a given rough rule set.The first step is to formulate a suitable coding system such as ordinal number system and the rough rule set then can be coded by the system to create a coded rough table.Each object of the coded rough table is processed to create an OSIT.The OSIT has to be further processed with regard to extract full information for the object.After the result of each OSIT for all objects was gained,they were organized to form extended table whose records are no less than the original rough set.But those redundant records can be directly excluded.The left records can not be excluded,otherwise the information of the original rough set will be lost.The least rule set is composed of those records.The last step is to recode ordinal number into the original meaning and the generation of the least rule set for a given rough set is completed.A computer program was written to test our analysis.In our experiment,we used several possible factors that may or may not have influence upon soil type,but we do not know which factors have or have not beforehand.If those unimportant factors are not excluded from important factors,low speed and efficiency will be inevitable when computer does intelligent reasoning.If the least rule set is stored in the knowledge,the analyzing speed can be greatly enhanced.We got the least rule set based upon OSIT analysis and processing.This method shows great prosperity in intelligence decision system in other fields.