#artworks {
# Adaptive code
# CSS code is also valid
}
Select nodes as in CSS
Select the nodes where the adaptation should take place as you would do with CSS.
@user (min-art-knowledge: 50) {
@context (max-hour: 18.00) {
#artworks {
# Adaptive code
# CSS code is also valid
}
}
}
Target users and contexts
Target specific users and / or contexts so the adaptation.
@user (min-art-knowledge: 50) {
@context (max-hour: 18.00) {
#artworks {
reorder-nodes: data-order;
color: red;
}
}
}
Select the adaptation effect
Select the nodes where the adaptation should take place as you would do with CSS.
-
Distributed User Model
User model distributed between server and client browser. Based on HTML5 standards so no special software is needed. Sensitive data is kept private.
-
Balancing Privacy and Personalization
Users can freely send their data to the server. In exchange some personalization techniques such as collaborative filtering are anabled for them.
-
Transparent and Flexible
A user-friendly interface is provided so the user is aware of all the information collected about her and can make decisions about that information.
What is this about?
The world wide web is an enormous hyperspace where users face the problem of information overload. Adaptive web based systems try to tackle this problem by displaying only the information that is really meaningful for the user. These systems need to collect data from the user in order to personalize the information. The set of information that the system has collected about a user is called the User Model.
User models in adaptive web base systems are typically stored on the server. However, this has some issues such as lack of privacy, server overload, band-width usage, limitation of events that can be tracked, lack of context awareness, etc... To solve this problem, some client side approaches have also been proposed. Still, client based user modeling has some other drawbacks. Typically the user has to install some piece of software, like a desktop application or a browser plugin, and techniques that rely on the comparison of several user profiles cannot be applied. P2P networks allow the analysis of several client user profiles at a time, but in that case the result will depend on the peers connected at the moment when the comparison is being performed.
WiBAF aims to balance these two approaches in a way that the advantages of both are maximized and the drawbacks minimized.