并举例说明，对于不一致决策表，其属性约简的代数表示不能用条件信息量来等价表示。
Through examples, it shows that attribute reduction of an inconsistent decision table cannot entirely be represented by conditional information quantity.
属性约简要求在保持知识库的分类和决策能力不变的条件下，删除不相关或不重要的属性。
Under the condition of unchanged classification and decision abilities, attribute reduction is to delete irrelative or unimportant attribute.
提出了一种基于条件粒度熵的属性约简的启发式算法，通过例子分析，表明该算法是有效的。
The paper offer a new heuristic attribute reduction algorithm based on conditional granularity entropy, though running an example, we show that this algorithm is effective.
应用推荐