Abstract: Objects classification is strict excessively and too sensitive on noise. Aiming at decision system with uncertain factor, an algorithm of attribute reduction based on dependability is established. The attributes in the system with uncertain information and noise data are reduced, whereby finding a reducible implied pattern of the data, and deleting those redundant rules in the system. And it can keep the original properties and functions of the system. The implementation of algorithm is described by introducing an example.

Key words: rough set, dependability, attribute reduction, implementation

摘要： 传统粗糙集分类方法过于严格，对噪音过分敏感。针对带不确定因子决策系统，提出一种基于属性依赖度的约简算法，使含不确定信息及数据噪音的系统中的属性得以简化，找到一种具有广泛表达能力的数据隐含格式，删去冗余的规则，并保持系统的原有用途和性能。通过一个例子实现了该算法。

关键词: 粗糙集, 依赖度, 属性约简