Exploitation of a crisp relation in a ranking problem

The outranking methods usually contain two main steps: determination of a (sometimes valued) relation which aggregates the preferences and exploitation of this relation in order to choose, sort or rank the actions. In a ranking problem, the second step is the construction of a (complete or partial) preorder from a (valued) relation. We have considered it was interesting to list the good properties such a construction should verify and to analyze some procedures in this context. We present here the first results of this study, which concern the construction of a complete preorder from a non valued relation.