Study On The Approaches For Dynamically Updating Approximations Based On Variable Precision Rough Sets

Rough set theory is a kind of mathematical tools for dealing with and analyzing uncertain and fuzzy knowledge. It has been successfully applied in pattern recognition, expert systems, fault diagnosis, decision analyses and other aspects. With the variation of data in database, the original knowledge in database can't satisfy people's demands. Therefore, it is meaningful to study how to efficiently obtain knowledge according to the original information in database and better support our decision making.When the attribute dynamically changes in information systems, the traditional approach for updating approximations is re-division of the universe. It costs a lot of recalculating time. Here, an approach which avoids re-division of the universe is proposed. The efficiency of dynamically updating approximations is improved. By analyzing the relationship between equivalent classes and original approximations, the corresponding theorems between updated approximations and original approximations are given. Then, the approaches for dynamically updating approximations while adding or deleting an attribute are respectively proposed in classical rough set model. The experimental results verify the validity of the approaches and the efficiency of the proposed approaches are better than that of the original approach.In classical rough set theory, the data must be accurate, namely, there is no noise data or data with a missing value. However, in real applications, there are many reasons which may lead to the existence of noise or incomplete data. Therefore, the variable precision rough set model was proposed by Ziarko to aim at modelling classification problems involving uncertain or imprecise information. In the variable precision rough set model, by studying on the changes of equivalent classes in information systems, several theorems and corollaries are given. Then, the approaches for dynamically updating approximations are respectively proposed in the variable precision rough set model while adding or deleting an attribute. Experimental results show the validity of the proposed approaches.