A survey on non-conventional preference modeling

In the paper a survey on non-conventional preference modeling is presented. Some philosophical discussion on the logic and foundations of human preferences is presented as the first attempt to provide an axiomatization of the subject. The use of preference models in decision aid however introduces such parameters to put serious questions to the possibility to build a complete normative axiomatization of this discipline. The conventional preference models are discussed both from a mathematical and the decision aid point of view underlining the limitations and impossibilities of this approach towards, complex and particular problem situations. The extensions of the conventional theory, while enhancing its potentialities, fail to give a satisfactory answer to such problems as uncertainty, ambiguity and imprecision of preference models, mainly when multidimensional preferences have to be considered (collective and/or multicriteria preferences). Two non-conventional approaches are presented: fuzzy preference models and the concept of outranking relation. In the first case an account of the theoretical foundations of valued preference relations is presented and some problems with axiomatization are discussed. In the second case, while the operational potentialities of the approach are underlined, the lack of a strong theoretical foundation is emphasized. A new four-valued logic is considered as a tool enabling, to give such a theoretical support. The future research on the subject concludes the paper.