属性约简中的近似问题研究

• 作 者: 高黎 沙基昌 余滨 卜先锦
• 机 构: 国防科技大学信息系统与管理学院,国防科技大学信息系统与管理学院,国防科技大学信息系统与管理学院,国防科技大学信息系统与管理学院 长沙410073,长沙410073,长沙410073,长沙410073
• 出 处: 系统工程理论与实践
• 发表日期: 2007年02期
• 关键词: 粗糙集 属性约简 相对约简 近似相对约简

现有相对约简实际上只是在保持所关心范畴的下近似不变的前提下对属性进行约简,这对于不协调信息系统来说,可能会使所关心范畴的上近似变大,从而使其近似精度减小,导致有用信息的丢失.为此,针对属性约简中的近似问题进行了研究,引入近似相对约简概念,利用近似相对约简替代现有相对约简来处理不协调信息系统,以克服现有相对约简在不协调信息系统中所带来的缺陷.同时,证明了在协调信息系统下,条件属性集相对决策属性集的近似相对约简和相对约简是等价的,并提出了一种基于近似相对约简的简便属性约简算法.

The existing relative reductions actually reduce the attributes only while maintaining the similarities and differences under the category of interest, which may cause the approximation of the categories of concern to become larger for uncoordinated information systems so that Its approximate accuracy is reduced, resulting in the loss of useful information.Therefore, the approximate problem in attribute reduction is studied, the concept of approximate relative reduction is introduced, and the approximate relative reduction is used to replace the existing relative reduction to deal with the uncoordinated information System to overcome the shortcomings of the existing relative reduction in the uncoordinated information system.At the same time, it is proved that the relative reduction and relative reduction of the relative decision attribute set of the conditional attribute set are equivalent under the coordinated information system , And a simple attribute reduction algorithm based on approximate relative reduction is proposed.