动态量化非对称相似关系的不完备知识约简算法

Incomplete Knowledge Reduction Based on Dynamic Valued Asymmetric Similarity Relation

量化非对称相似关系是处理不完备信息系统的重要工具之一.本文针对非对称关系中明显相似的对象分类不合理的问题,定义了动态量化非对称相似关系,提出带有自动阈值调节的动态量化非对称相似关系模型,根据实际数据自动确定其阈值,使之更加灵活和合理.并采用快速排序提高知识约简过程中相容类的计算效率.通过实例验证了该算法处理不完备知识约简的有效性.最后,应用该模型解决了地下空间信息化施工的不完备知识约简问题.

The valued asymmetric similarity relation plays an important role in dealing with incomplete information systems. Since it is unreasonable for asymmetric similarity relation to classify obviously similar relation, dynamic valued asymmetric similarity relation is defined and a kind of valued asymmetric similarity relation model with auto-selecting threshold is built. This model makes data se- lection more flexible and reasonable. The quick sorting is applied to compute consistent class in order to improve reduction efficiency. The algorithm＇s effectiveness is proved to process incomplete knowledge reduction. Finally, this algorithm is employed to reduce in- complete knowledge of information construction for underground space.