Hyperlink Recommender Systems Design

The information age is characterized by an§overabundance of information but few tools to help§users of information spaces get just the right§information as they navigate these spaces. Current§research thus aims to design tools like recommender§systems to personalize the user experience in large§information spaces. This book explores factors§involved in hyperlink recommender systems design with§a view to improving on them or proposing new§solutions. The work makes the following §contributions in the area of data mining.§Web Page Classification: a new classification metric§is developed; Association rule mining: a new apriori§algorithm is developed; Prediction model for user§interests: methodology for extracting from web§server logs, association rules that show§correlations between user navigation patterns and§interesting web pages, and transformation of the§rules into collaborative filtering data, widely used§in recommender systems; Clustering: a comparative§study of the CLARANS, PAM, and CLARA algorithms in§high dimensional space is presented. The work should§therefore be of interest to anyone involved in§personalization systems and data mining research.