Study On The Approaches For Dynamically Updating Approximations Based On Covering Rough Sets

There are a lot of research achievements about incremental updating methods of approximation sets under the classic rough set model and extended rough set models with the attribute set in information systems varies. The ultimate goal of incremental updating research of approximation set under all sorts of rough set models is how to efficiently obtain dynamic rules using the existing knowledge. Based on the existing achievements using the relation between the equivalence relation and approximation set and rule sets, this paper proposes a method for incremental updating decision rule sets. The effectiveness of the proposed method is valified through examples.When the attribute set varies, the traditional method of updating approximation of covering rough sets is to compute the covering elements repeatly, and make of the cover of the unverise. Then, all the covers merge together to constitute a new cover family. It needs a lot of time to repeatly compute neighborhoods of all the objects. This paper discusses the variation trend of neighborhoods of objects when an attribute is added or deleted respectively. Then the relational theorem between the updating approximation sets and original approximation is given through discussing the relation between neighborhoods of objects and original approximation sets. Finally, a method for dynamically updating approximation sets with the attribute set varies is proposed. The experimental results verified the validation of this method and it is more efficient than the original method.So far, there are some Rough set tool softwares, such as Rosetta, ROSE, Rough Enough and LERS. But these system or program libraries were developed based on the classic rough set theory. A software tool for knowledge discovery based on covering rough sets is developed. It has high maneuverability and with object-oriented characteristics.