Use of fuzzy rough set attribute reduction in high scent web page recommendations

Abstract

Information on the web is growing at a rapid pace and to satisfy the information need of the user on the web is a big challenge. Search engines are the major breakthrough in the field of Information Retrieval on the web. Research has been done in literature to use the Information Scent in Query session mining to generate the web page recommendations. Low computational efficiency and classification accuracy are the main problems that are faced due to high dimensionality of keyword vector of query sessions used for web page recommendation. This paper presents the use of Fuzzy Rough Set Attribute Reduction to reduce the high dimensionality of keyword vectors for the improvement in classification accuracy and computational efficiency associated with processing of input queries. Experimental results confirm the improvement in the precision of search results conducted on the data extracted from the Web History of "Google" search engine. © 2009 Springer-Verlag Berlin Heidelberg.