This paper presents most used user models and inductive methods based on probability. It describes a typical usage of user modelling in a complex recommender system and propose a new contribution to probabilistic user modelling.
Some initial experiments are also presented.