How deriving benefits from expert advices to make the right choice in multi-criteria decisions based on the Choquet integral?
Résumé
The Internet and World Wide Web have brought us into a world of endless possibilities. It could appear as a genuine benefit but this world also is one of endless choice: how can we select from a huge universe of items of widely varying quality? How can we take into account many evaluation criteria to soundly evaluate all these potentials choices? We need expert advices. Recommender Systems (RS) have emerged to address this issue: they're based on the customers' critics and their authentic experiences upon the competing items. The purpose of RS is thus providing recommendations from people who are familiar with the choices we face: they can be considered as experts from this point of view. The example we deal with in this paper is the choice of an e-business site (e-retailer) when making purchases on the web. On the basis of the gathered expert critics, this paper explains how to design multi criteria evaluation aids and traceability functions for legitimate choices and logical argumentation. The evaluation of each competing e-retailer results from a multi criteria assessment by the collected experts' critics. A Choquet integral based aggregation is used to consider interactions between criteria in the overall user rating. The argumentation principle underlying this approach is based naturally on coupling the base of expert critics in natural language and an information fusion system. The reliability of the ranking introduces the notion of decisional risk. All these proposals are extended in the case of imprecise or uncertain evaluations by experts.