粗糙集理论中新的针对不完备信息系统的处理方法研究

本文针对不完备信息系统,在分析了现有的数据补齐法和扩充法的优劣后,提出基于差异关系和数据部分补齐的处理方法。本文定义了差异关系,对差异矩阵进行扩充使其能适用于不完备信息系统,证明了用差异关系进行属性约简和求核的可行性,并给出了相应的算法。数据部分补齐的目的是通过分析对象之间的差异性更进一步地挖掘信息之间的潜在联系,为后续的决策规则的求取提供更丰富更准确的信息。实验证明,在处理不完备信息系统时,基于差异关系和数据部分补齐的方法能获得更好的分类性能。

Research on Rough Sets under Incomplete Information System

In order to process incomplete information systems in Rough sets, a novel method based on discrepancy relation and data partial supplement is introduced. This paper analyzes current typical methods, defines the discrepancy relation, extends the discrepancy matrix, proves the feasibility of using discrepancy relation to reduce redundant attributes and provides corresponding algorithms. Besides, uses a new method of data partial supplement to obtain latent data to provide more accurate information for next step of extracting decision rules. The experiment results show that the new method can get best performance evaluation.