DocumentCode :
1967507
Title :
Hybrid approach for predicting the behavior of Web users
Author :
Kao, Darren Ming-Shan ; Özyer, Tansel ; Alhajj, Reda
Author_Institution :
Dept. of Comput. Sci., Calgary Univ., Alta., Canada
fYear :
2005
fDate :
15-17 Aug. 2005
Firstpage :
217
Lastpage :
222
Abstract :
Hybrid approach for predicting the behavior of Web users in this paper, we propose the design and implementation of a hybrid system by combining several data mining techniques to capture user\´s Web browsing behavior. User navigation sessions that represent the interaction with a given Website are used to construct hypertext probability grammar (HPG). The production with high probability in HPG represents the most favorable user browsing trail. The HPG results will be further used to construct N×M matrix, and a clustering algorithm are applied to extract clusters of behaviors. N-gram model is used based on the assumption that Website visitors have limited memory of what they visited before, and the choice of the next page to visit does not depend on all pages visited previously; but only the N -1 page. N-gram will not generate strong x where |x| \n\n\t\t
Keywords :
Internet; data mining; hypermedia; matrix algebra; online front-ends; N-gram model; Web browsing behavior; Web user behavior; clustering algorithm; data mining; hybrid system; hypertext probability grammar; matrix compression; semantic constraint; user profiling; visual representation; Clustering algorithms; Computer science; Data mining; History; Information analysis; Navigation; Pattern analysis; Production; Web mining; Web server;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Information Reuse and Integration, Conf, 2005. IRI -2005 IEEE International Conference on.
Print_ISBN :
0-7803-9093-8
Type :
conf
DOI :
10.1109/IRI-05.2005.1506476
Filename :
1506476
Link To Document :
بازگشت