An algorithm of Web usage mining based on attributes number set

As large session pattern are saved in Web server log, aiming at the character that these pages attribute is Boolean quantity, this paper proposes an algorithm of Web usage mining based on attributes number set, which is suitable for mining session pattern containing many visiting pages. The algorithm turns session pattern of users into binary, and then uses the way of number descending to generate frequent candidate item sets. The algorithm computes support by attributes number set dimensions in order to scan once session pattern of users, and then the efficiency of Web usage mining is efficient improved. The experiment indicates that the efficiency is faster and more efficient than presented algorithms. © 2010 IEEE.