Skip to Main content Skip to Navigation

Une approche épiphyte pour la conception de systèmes conseillers

Abstract : Recommender systems are systems that support the use of softwares. In the context of the Web recommender systems are widely used to facilitate (or extend) search functionalities or assist users in their navigations through Websites. In this thesis we propose an approach to build a generic recommender system based on a proxy-like (epiphyte) architecture. The recommender system intercepts the users' requests and the given Website answers, analyzes them and proposes tips. This particular architecture allows to separate the data of the target Website from the one of the recommender system, facilitating maintenance and evolution of the both systems. We specifically target complex Websites, i.e., websites that have a large and in-depth structure, numerous pages and functionalities, and are meant to be used for different purposes. The help proposed is guidance and is based on the task the user want to realize when he accedes to the Website. To offer such an help, the recommender system analyzes the user's navigation according to models of navigation and builds tips on the fly. Models of navigation are built by experts and embed the tips that are to propose to users. They represent prototypical uses of the Website, i.e., different ways to browse the Website and to reach a goal. As the model of navigation building task appears to be difficult, we propose two computer approaches to assist experts, the second one offering more advanced tools. Two prototypes have been built. They have permitted us to illustrate our approach and can be considered as proof of concept.
Complete list of metadatas

Cited literature [22 references]  Display  Hide  Download
Contributor : Bruno Richard <>
Submitted on : Monday, June 14, 2010 - 11:12:33 AM
Last modification on : Friday, October 23, 2020 - 4:42:44 PM
Long-term archiving on: : Tuesday, September 14, 2010 - 8:26:08 PM


  • HAL Id : tel-00491735, version 1


Bruno Richard. Une approche épiphyte pour la conception de systèmes conseillers. Autre [cs.OH]. Université du Maine, 2008. Français. ⟨tel-00491735⟩



Record views


Files downloads