Skip to Main content Skip to Navigation
Journal articles

Modeling users' search through contextual graphs.

Abstract : This paper presents a work on a comparison between a user model and user's behavior based on three premises. First, any system includes a representation of its users. Second, the external representation of users in a system is related to how the system is used by users. Third, knowing how to use the system depends on the task context. For making context explicit in order to use it, we use contextual graphs to capture the effective behaviors of users in an activity of information retrieval on a scientific website. We show how such a context-based representation may help to predict the search behavior of visitors to the website pages. We extend the capability of the system by a preliminary study based on a technique of eye tracking coupled with the contextual-graph representation. This approach allows dealing with a system that is able to incrementally acquire new knowledge from the user and learn new practices when the system is in a situation of failure.
Document type :
Journal articles
Complete list of metadatas
Contributor : Charles Tijus <>
Submitted on : Tuesday, October 11, 2011 - 12:41:28 AM
Last modification on : Friday, January 8, 2021 - 5:32:06 PM

Links full text



Patrick Brézillon, Laure Léger-Chorki, Charles Tijus. Modeling users' search through contextual graphs.. Revue des Sciences et Technologies de l'Information - Série RIA : Revue d'Intelligence Artificielle, Lavoisier, 2009, 23 (4), pp.467-484. ⟨10.3166/ria.23.467-484⟩. ⟨halshs-00630785⟩



Record views