Abstract: The problem in the usual reduction method of a decision table by means of Rough set theory is that it does not make full use of the information given by discernible matrix so that it gives rise to the repeated computation in computing the core value and the value reduction of the decision rules. The present paper aims at improving the present value reduction strategy used widely, which can get the core values of each rule from the discernible matrix. Then the DMBVR algorithm is given, which can get all the value reductions based on the discernible matrix. Finally it is proved that DMBVR algorithm can get all the reductions of each rule and makes the process of computing the core value and the value reduction more simple.

Key words: Rough set, discernible matrix, core value, value reduction