Séminaire (organisé par l’équipe de recherche ICI)

Sébastien Destercke

Chargé de Recherches CNRS, Heudiasyc

Learning to rank instances through questions : some basic aspects

Mardi 26 avril 2016 à 14 h en GI 041

Résumé :

Ranking instances (objects, profiles, people) is a problem that happens in a lot of situations : paper reviewing, applicant selection, recommendation systems. While this problem is an old one, the interest for it has been renewed for with the advent of online recommendation systems. In this talk, we will interest ourselves to two issues and their related research questions : identifying a ranking model, and reducing uncertainty about an instance to better identify its interest. We will consider the case of active and personalized questions to the decision maker, rather than the case of automatically learning a model from passively collected information. The various issues will be illustrated on a simple model


FR SHIC 3272

Collegium UTC/CNRS