论文部分内容阅读
Decision table is an important tool for knowledge acquisition in rough set theory. In order to reduce the fuzziness and randomness of the decision table with continuous attributes, a novel reduction algorithm based on grey relational degree is proposed. Firstly, the decision table is converted to the same domain. Then, the grey relational matrix is constructed to describe the equivalence relations between samples. Finally, an improving dynamic clustering method is adopted to extract the coarser granularity of the sample, which can automatically cluster the similar samples. The experiment shows that the reduction decision table has gotten roughly the same recognition rate with less than one-tenth the size of the original condition. Thereby it significantly reduces the knowledge acquisition time for rough set.