An Intelligent Decision Support System Based on Machine Learning and Dynamic Track of Psychological Evaluation Criterion

An Intelligent Decision Support System based on Machine Learning and Dynamic Track of Psychological Evaluation Criterion is presented in this paper. It is shown that a complex decision for global situation can be disassembled into a series of simple local problems, from which the most satisfactory decision for the local can be found out by individual ways respectively. At the lower level of total score, the best decision for the local, according to the mathematical interpretation of weight, can be considered as the decision whose distribution of scores is just consistent with the distribution of the psychological weight (or preference) of a decision maker. At a series of moderate levels, the evaluation criterion is given by human–machine interaction, in which some satisfactory samples are chosen by decision maker from a lot of samples, and the barycentre of criterion and the radius of criterion can be estimated by a learning algorithm. In this way, the most satisfactory decision for the local made by the decision maker at each level can be tracked. If we let the collection of satisfactory decision for global be the union of the local’s most satisfactory decision at all levels, then the changing process of psychological criteria which varies with the change of total score can be deduced. Finally, a satisfactory degree function with which the global consistency of the collection of local satisfactory decisions at all levels could be retained is given, and a global ranking approach based on the function as well.