Skip to Main content Skip to Navigation

Un cadre possibiliste pour l'aide à la décision multicritère et multi-acteurs - Application au marketing et au benchmarking de sites e-commerce

Abstract : This thesis is situated within the framework of a multi criteria and multi-actors decision making process. It concerns the management of certain aspects of the uncertainty inherent to the multi criteria evaluation in such a process. We propose a modelling and a mathematical formalization of this process via a possibilistic representation of the evaluations. We also suggested characterizing the possibility distributions relative to the evaluations of the various criteria by a set of description indicators such as the imprecision, the divergence, etc., so allowing an easier quantitative interpretation for the decisionmaker. We considered aggregation operators of the family of the Choquet integral to propagate the distributions relative to criteria evaluations as well as the description indicators of distribution, because of their capacity to take into account interactions between criteria while keeping a character of "generalized" average easily understandable by the decision-maker. Concerning the propagation of indicators, we established two key theorems, the one concerning the union of the superior and lower values of the mean interval of a possibility distribution and the other one the mean imprecision of a possibility distribution. Besides, the concept of explanation of the decision particularly constituted the kernel of our model of decision making process. So from the established theorems, we proposed explanation features based on a quantification of the contribution of decision-making indicators of the criteria evaluations to the global evaluation. We so showed that each indicator concerning the global evaluation distribution is simply written as a weighted sum of criteria contributions. As application software, we were interested in the formalization of the process of management of customers' satisfaction measures in the context of the e-recommendation for the e-commerce. We presented a method of multi criteria valuation of the e-commerce sites by Internet users' community which allows at the same moment to take into account the relative importance of the evaluation criteria and the interactions between these, but also to manage various aspects of the uncertainty inherent to this process of evaluation. We developed a software prototype serving as decision support system. This system named "Feedback Based Recommendation System" (FBRS) allows to supply the managers of an e-commerce site, in particular the marketing and benchmarking teams, elements which allow them to understand the scores which were attributed to their site during the evaluation by the customers. We showed how it is then possible to identify the most decisive criteria in this evaluation and how to use this contributions analysis in the monitoring and management of the performances by the marketing and benchmarking managers. These features are a working base for them to improve the existing strategies.
Document type :
Complete list of metadatas

Cited literature [58 references]  Display  Hide  Download
Contributor : Afef Denguir <>
Submitted on : Wednesday, February 20, 2008 - 7:23:28 PM
Last modification on : Friday, November 6, 2020 - 3:26:37 AM
Long-term archiving on: : Thursday, May 20, 2010 - 11:09:54 PM


  • HAL Id : tel-00258010, version 1


Afef Denguir. Un cadre possibiliste pour l'aide à la décision multicritère et multi-acteurs - Application au marketing et au benchmarking de sites e-commerce. Autre [cs.OH]. Université de Savoie, 2007. Français. ⟨tel-00258010⟩



Record views


Files downloads