Link
Corin R. Anderson
Pedro Domingos
Daniel S. Weld
International World Wide Web Conference archive
Proceedings of the tenth international conference on World Wide Web
Hong Kong, Hong Kong
Pages: 565 - 575
Year of Publication: 2001
ISBN:1-58113-348-0
Excerpt of Abstract: To best serve the needs of this growing community, we propose building web site personalizers that observe the behavior of web visitors and automatically customize and adapt web sites for each individual mobile visitor. In this paper, we lay the theoretical foundations for web site personalization, discuss our implementation of the web site personalizer PROTEUS, and present experiments evaluating its behavior on a number of academic and commercial web sites. Our initial results indicate that automatically adapting web content for mobile visitors saves a considerable amount of time and effort when seeking information "on the go".
Keywords: Adaptive web sites, personalization, wireless web
My Discussion:
Describing personalizing a web site by using mathematical notation certainly looks rigorous, but the end result is program that seraches through the space of all possible permutations of a site (removing / adding / transposing links in pages and content) by applying a fitness function to each possible personalization of the site. As the authros admit, the optimally personalized site can't be computed this way without significantly pruning the search space down, and one of the ways they do this is by having the resulting site always give more precedence to pages the user is sure to reach, like front page. The model of the user's needs thus ends up skewed towards what the user is bound to reach, not what the user wants. The user's needs are modelled based on where the user went before (probably when using browsers on the big screen), and the function does, IMHO correctly, assign a far higher likelyhood that the user will scroll than that the user will retrieve a link. No metnioning of the problems in tracking users between devices, and that users will have to somehow identify themselves on every device to the site. Their graph of results shows that while in 9 out of ten information retrieval tasks their test users required fewer clicks to get to the desired data (unpersonalized site was on a desktop, personalized site was navigated on a Palm), the time they spent on the sites was equally long for the task. No word on user perception.
Corin R. Anderson
Pedro Domingos
Daniel S. Weld
International World Wide Web Conference archive
Proceedings of the tenth international conference on World Wide Web
Hong Kong, Hong Kong
Pages: 565 - 575
Year of Publication: 2001
ISBN:1-58113-348-0
Excerpt of Abstract: To best serve the needs of this growing community, we propose building web site personalizers that observe the behavior of web visitors and automatically customize and adapt web sites for each individual mobile visitor. In this paper, we lay the theoretical foundations for web site personalization, discuss our implementation of the web site personalizer PROTEUS, and present experiments evaluating its behavior on a number of academic and commercial web sites. Our initial results indicate that automatically adapting web content for mobile visitors saves a considerable amount of time and effort when seeking information "on the go".
Keywords: Adaptive web sites, personalization, wireless web
My Discussion:
Describing personalizing a web site by using mathematical notation certainly looks rigorous, but the end result is program that seraches through the space of all possible permutations of a site (removing / adding / transposing links in pages and content) by applying a fitness function to each possible personalization of the site. As the authros admit, the optimally personalized site can't be computed this way without significantly pruning the search space down, and one of the ways they do this is by having the resulting site always give more precedence to pages the user is sure to reach, like front page. The model of the user's needs thus ends up skewed towards what the user is bound to reach, not what the user wants. The user's needs are modelled based on where the user went before (probably when using browsers on the big screen), and the function does, IMHO correctly, assign a far higher likelyhood that the user will scroll than that the user will retrieve a link. No metnioning of the problems in tracking users between devices, and that users will have to somehow identify themselves on every device to the site. Their graph of results shows that while in 9 out of ten information retrieval tasks their test users required fewer clicks to get to the desired data (unpersonalized site was on a desktop, personalized site was navigated on a Palm), the time they spent on the sites was equally long for the task. No word on user perception.
# posted by Unknown @ 5:00 PM
Comments:
Post a Comment
