Laboratoire de Génie Informatique et d’Automatique de l’Artois

In this document, the conception of decision support systems in an imperfect context is approached. The goal of these systems is to help the expert to make a decision in the case of complex systems. A complex system is a system for which the decision is based on a large number of information. Moreover, these information can be imperfect (imprecise, uncertain, incomplete, etc.).

To realize decision support systems, three steps are needed:

1. the data acquisition,
2.  the modelling the information,
3. the knowledge extraction (or make the sense) to provide help to the expert for making a decision.

In this habilitation thesis, two last steps of this process are explained. To deal with the imperfect information, we use the belief function theory considered as the unifying formalism in the uncertainty theories. This theory allows us to easily represent the knowledge. Furthermore, the conception of two decision support systems is introduced in this manuscript. The first application concerns the selection of supplier in the framework of supply chain management, whereas second one deals with the decision support for forensic investigation.