关键词: 粗糙集，不完备信息系统，极大相容块，数据填补，决策树

Abstract: Maximal consistent block technictue has some advantages in dealing with the incomplete information. But the construction of maximal consistent blocks itself can be a timcconsuming process. After the properties of maximal consistent blocks in incomplete information system were analyzed, the consistent data reinforcement method was defined to handle the missing data in maximal consistent blocks. Using the method in incomplete information system, a new algorithm was obtained to construct maximal consistent block. And the decision tree structure was introduced to optimize the algorithm based on the algorithm' s characteristics. Algorithm was evaluated on the standard benchmark dataset.The result of the experiments indicates that the algorithm has better performance in large scale dataset.

Key words: Rough set, Incomplete information system, Maximal consistent block, Data reinforcement, Decision tree