【摘要】：正The knowledge reduction methods(Pawlak 1991) allow us to find core and keys for condition attributes in a decision table.For each key,the decision table is reduced to a key decision table in which the key becomes the set of condition attributes.To discover knowledge,we need to make knowledge reasoning for every tuple in the key decision table.That is,we need to find core values and key values for each tuple.We present a series of algorithms to do this knowledge reasoning and illustrate our methods by an optician's decision table.