Research Of Attribute Reduction Algorithm Based On Rough Set、 T-norm And Evidence Theory

Nowadays, with the rapid development of the Internet, the data we can get becomes more and more complex and there be more and more redundant data. Thus, do attribute reduction is essential and through reduction properties can be play a very good guide to the huge data we are dealing with. This article is a study of attribute reduction from there different perspectives.First, it is a very important research topic to study the attribute reduction in rough set.In front of a lot of information, only part of the information plays a decisive role in actual problems and some other information can be deleted directly. Removing redundant information helps us find even more decisive attributes. The most classic and traditional attribute reduction algorithm is based on Pawlak reduction. In rough set theory, it deals with the uncertainty of system by introducing approximation precision()Rd X. However, the traditional division of precision can not distinguish between a good degree of fine particles. Firstly, it scored divided by the fineness of the particles that takes into account the approximate division of proportion to this ratio and the roughness of the product to modify traditional approximation accuracy, which can ensure that the knowledge divide finer, the greater the accuracy. At the same time the accuracy of the new definition is applied to the attribute reduction, and by example to prove the accuracy of attribute reduction must be Pawlak reduction.Secondly, triangular norm has many excellent properties such as commutative, associative and monotonic. Therefore, from this direction to consider the proposed attribute reduction it should also have a lot of excellent properties. Firstly, by constructing a T- norm, by this T- norm to characterize the similarity between two sets, then the similarity extended to measure the similarity between equivalence classes. Through examples demonstrate the effectiveness of this method. This method is a innovative method in this article.Finally, put forward the idea of a deal with the continuity of data. In dealing with the continuity of data, the first by evidence of theoretical knowledge to successive discrete data, and then use the rough set knowledge of discrete data attribute reduction, and finally simultaneous the importance of attribute in rough set and knowledge of evidence to make decisions for continuous value problem. Effective integration of Rough Sets and Evidence Theory can be more objective and more scientific process data, which can effectively find the most important attributes when dealing with continuous data can play a very good effect.