High Scent Web Page Recommendations Using Fuzzy Rough Set Attribute Reduction

Abstract

AbstractInformation on the web is huge in size and to find the relevant information according to the information need of the user is a big challenge. Information scent of the clicked pages of the past query sessions has been used in the literature to generate web page recommendations for satisfying the information need of the current user. High scent information retrieval works on the bedrock of keyword vector of query sessions clustered using information scent. The dimensionality of the keyword vector is very high which affects the classification accuracy and computational efficiency associated with the processing of input queries and ultimately affects the precision of information retrieval. All the keywords in the keyword vector are not equally important for identifying the varied and differing information needs represented by clusters. Fuzzy Rough Set Attribute Reduction (FRSAR) has been applied in the presented work to reduce the high dimensionality of the keyword vector to obtain reduced relevant keywords resulting in improvement in space and time complexities. The effectiveness of fuzzy rough approach for high scent web page recommendations in information retrieval is verified with the experimental study conducted on the data extracted from the web history of Google search engine.KeywordsFuzzy rough setInformation retrievalInformation needInformation scentFuzzy similarity