ABSTRACT. In a large number of problems such as product configuration, automated recommendation and combinatorial auctions, decisions stem from agents subjective preferences and requirements in the presence of uncertainty. In this paper, we introduce a framework based on the belief function theory to deal with problems in group decision settings where the preferences of the agents may be uncertain, imprecise, incomplete, conflicting and possibly distorted. A case study is conducted on product recommendation to illustrate the applicability and validity of the proposed framework.