Discernibility System in Rough Sets.

Zongtian Liu,Zhipeng Xie
DOI: https://doi.org/10.1007/3-540-48912-6_30
1999-01-01
Abstract:In the classic rough set theory [1], two important concepts (reduct of information system and relative reduct of decision table) are defined. They play important roles in the KDD system based on rough sets, and can be used to remove the irrelevant or redundant attributes from practical database to improve the efficiency of rule extraction and the performance of the rules mined. Many researchers have provided some reduct-computing algorithms. But most of them are designed for static database; hence they don't have the incremental learning capability. The paper first proposes the idea of discernibility system and gives out its formal definition, then presents the concept of reduct in discernibility system, which can be viewed as a generalization of the relative reduct of decision table and the reduct of information system. At last, based on the concept of discernibility system, an incremental algorithm, ASRAI, for computing relative reduct of decision table is presented.
What problem does this paper attempt to address?