Fuzzy–rough attribute reduction with application to web categorization

Due to the explosive growth of electronically stored information, automatic methods must be developed to aid users in maintaining and using this abundance of information effectively. In particular, the sheer volume of redundancy present must be dealt with, leaving only the information-rich data to be processed. This paper presents a novel approach, based on an integrated use of fuzzy and rough set theories, to greatly reduce this data redundancy. Formal concepts of fuzzy–rough attribute reduction are introduced and illustrated with a simple example. The work is applied to the problem of web categorization, considerably reducing dimensionality with minimal loss of information. Experimental results show that fuzzy–rough reduction is more powerful than the conventional rough set-based approach. Classifiers that use a lower dimensional set of attributes which are retained by fuzzy–rough reduction outperform those that employ more attributes returned by the existing crisp rough reduction method.