Résumé en anglais

This paper deals with preference modelling in the context of Decision Aid. In this framework, conflicting systems of logic, uncertain knowledge, ambiguous positions are always present. In order to tackle this problem, a multiple criteria methodology is proposed, mainly based on fuzzy outranking relations introduced both at one-dimensional and multi-dimensional levels. Some properties of outranking relations are investigated. Such relations are then combined using fuzzy logical connectives to generate relational systems of fuzzy preferences that are shown to be very useful to reflect the vagueness of information in the various preference situations that may be considered in the modelling process.