关键词: 粗糙集, 属性约简, 增量式计算, 区分矩阵

Abstract: At present, many static algorithms of knowledge reduction based on decision table have been proposed, however, since the objects in the actual decision table are often changed, these static algorithms are computationally time-consuming. Aiming at the problem, the concept of simplified decision table is introduced, and it is proved that attribute reduction based on the simplified discernibility matrix is equivalent to that based on discernibility matrix. On this condition, an efficient incremental computing algorithm for attribute reduction based on decision table is designed. Example results illustrate the efficiency and feasibility of the new algorithm.

Key words: rough set, attribute reduction, incremental computing, discernibility matrix