Decision-making is the crucial step involved in many human activities. Thepreferences of decision-makers among alternatives are usually in partial orderings dueto the complexity of real decision-making environment and lack of knowledge. Partialordering structure provides a natural, effective information representation andreasoning for the complexity of decision-making problems under uncertainty. Whilethere is a lot of qualitative information in the actual decision-making processes, anddecision-makers are used to evaluating alternatives. Now, we need to consider thedegrees of credibility or belief associated with the preferences given by differentdecision makers, as well as other types of uncertainty, such as consistency level ofdifferent opinions. These uncertainties are actually always accompanying the processof every one issuing his preference, Hence, It’s particularly important to study thepartially ordered preferences with belief degrees for decision making.We propose in this thesis different approaches to solve different decision makingproblems. Firstly for alternatives ordering of the partially ordered preferences withbelief degrees for decision making. Based on the context probability, we derived anew method for aggregating the partially ordered preferences with belief degrees fordecision making. Using the new context Probability and ORDER-BY-PREFERENCEalgorithm to obtain the overall rank ordering. Secondly for decision making problemsof some alternative solutions can not be sorted or can not give preference todetermine. we use the belief structure for represent partially ordered preferencebetween alternatives associated with belief degrees. Evidential reasoning approach isthen applied to combine the preferential opinions from different decision makers intoa collective belief structure matrix. Based on this collective belief structure matrix,two overall rankings(total orders) between the considered alternatives are obtainedvia a distance-based aggregation method.The final ranking of the alternatives, isproduced by using the intersection of these two total orders based on ELECTRE IIIrules.At last, two simple examples are provided to illustrate these approaches. |