Abstract: At present，there is few researchers have investigated attribute reduction based on incomplete decision table.Incomplete decision table of attribute reduction based on information quantity is a new issue.In this kind of attribute reduction，computation of tolerance relation of the incomplete decision table is the main computation.At the same time，it is more difficult than computing equivalent relation of complete decision table.The time complexity of the existed algorithms of attribute reduction based on information quantity is almost O（|C|²|U|²）.To lower the time complexity，this paper first analyzes the shortcoming of those algorithms.And also provids an efficient algorithm for computing the tolerance class.Then using the above algorithm，this paper designs an efficient algorithm of attribute reduction based on information quantity.Its time complexity is O（|C|²|U|²）.

Key words: rough set, attribution reduction, incomplete decision table, information quantity, algorithm complexity

摘要： 目前，基于不完备决策表的属性约简研究较少。基于信息量的不完备决策表属性约简是一种新的属性约简。由于在该属性约简中，计算相容关系是最主要的计算，也比计算等价关系要难得多。基于信息量的不完备决策表的属性约简算法的时间复杂度一般为O（|C|²|U|²）。为降低其时间复杂度，首先分析了老算法的不足，然后给出了一个效率较好的计算相容类的算法。最后设计了一个新的基于信息量的不完备决策表的属性约简算法，其时间复杂度为O（|C|²|U|²）。

关键词: 粗糙集, 属性约简, 不完备决策表, 信息量, 算法复杂度