DATA ENRICHING BASED ON ROUGH SET THEORY

Data enriching is the attempt to transfer a incomprehensible database,which is made up of a large amount of data, into a concise and natural representa-tion which is more comprehensible for users. Several problems re1ated to data en-riching based on Rough Set (RS) are discussed in this paper. A reduction strategybased on discernibility matrix, which defines the attribute significance as the num-ber of the attribute appeared in discernibility matrix, is employed to analyze morethan 4O databases fr0m UCI repository. In accordance with the principle of RS, theenriched database is equivalent to the one given under the consistent decision con-straint. Such an enriching process is called consistent enriching. But there are somedatabases, if some instances are deleted from the database, higher evaporation rateof attribute and instance can be obtained. These instances are the so-called excep-tions and the enriching process of considering exception is called inconsistent en-riching. This paper presents a method, called modified discernibility matrix, tomake inconsistent enriching, it is applied to enrich databases with continuous at-tributes, the results demonstrate that it is very efficient.