Nowadays, a lot of web sites propose a ranking of recent movies to people looking for a recommendation in order to be helped in their choice. Generally, in these sites, the movies are ranked according to ratings of many criteria (scenario, actors, special effects, ..). In addition to a global synthetic evaluation, the users want complementary information to support their choices, especially when many movies have quite similar evaluations. Therefore, in this paper, we propose elucidation functionalities of the aggregated results as well as a definition of risk expressions of one movie recommendation based on a multi-criteria aggregation with the Choquet integral. Finally, a first approach is proposed to control the recommendation risk by adding in the database new reviews about the movies placed nearly equal.