Decision table, consists of conditional attributes and decision attribute may be considered as a knowledge representation. Reduction of decision table plays important roles in the process of rule reduction. This paper discusses a process of rule reduction using an extended concept of logic minimization. First, this paper introduces a concept of weighted decision table which is generated from a relational data table. Here, the weighted decision table may have such situation that several similar values of conditional attributes may have correlation to different value of decision attribute with a certain weight. Therefore in the process of rules reduction, it is necessary to extend the concept of logic minimization in order to process such kind of decision table. An algorithm of rule reduction is proposed along with an extended merge operation. During discussion of the proposed concept, some illustrated examples are given to clearly understand the concept.
展开▼
