A modified algorithm of attribute reduction based on rough set was introduced in this paper. First. in order to overcome the in-completeness of the two common deftniton of attribute significance in traditional rough set, a new definition of attribute significance based on weighed sum was proposed. Then this definiton of attribute significance was applied to the heuristic value reduction algorithm, and it was also used to guide the problem of how to add attribute value after we obtained core attribute value in the decision table. Additionally, with this modified algorithm , in premise of ensuring redction effect , complete solution space can be acquired from the decision table with conflict re-cords. At last, it was used to knowledge base reduction of one expert system, the result showed its validity.%基于粗糙集理论提出一种改进的属性值约简算法;首先针对粗糙集中两种属性重要度定义存在的不准确性问题,提出一种加权平均的属性重要度定义;接着将此定义引入改进的属性值约简算法之中,在求得决策表中属性值核后,用它指导后续属性值的添加问题;改进后的算法可以在保证约简效果的前提下,从不相容决策表中获得完整的解空间；最后通过约简某型专家系统知识库的实例验证了所提算法的有效性.
展开▼
