Research On Variable Weights Operator With Induce Behavioral Functions

The variable weights (VW) operator is an important aggregation operator, which is applied to solve the comprehensive evaluation problems. The main feature of the VW operator is that aggregating values with variable weights depending on attibutes value vector.The research on the VW operator in comprehensive evaluation is reviewed. Combined with two practical evaluation issues, namely the U.S. highway speed limit problem and personalized moive recommendation problem, fistly, the research proposes a VW operator with attributes diminishing marginal substitution (VW-DMS operator). The VW-DMS operator has the function of inducing individuals to regard their all-round development. Secondly, the research provides a VW with referring to individual preferences (VW-RIP operator), which induces individuals to take the subjective initiative, mine cognitive potential, develop "open to suggestions" values.With regard to the limitations in the existing research, a series of research works are conducted as follows:(1) From the perspective of multi-attribute utility theory, giving the axiomatic definition of VW-DMS variable weights vector and state variable weights. VW-DMS abandon the traditional variable weighs standards based on homogeneous attribute values and set a new standards based on evaluator preferences.(2) Using individual preferences point measure individual preferences; Using constant weights vector and attributes value vector measure group preferences; Exploreing the link between individual preferences point and the positive ideal point; The research elaborates that individual preferences point will converge to the positive ideal point when the Individual score-sample size is sufficiently large.(3) Basing on individual preferences point, constant weights vector and attributes value vector, the research proposes a VM operator integrated individual and group preferences:VW-DMS. According to individual scoring sample set, providing a method of solving individual preferences point of VM-DMS. (4) Basing on VM-DMS, the research designs a personalized movie recommendation system in the background of YouTube movie recommendation. Through simulation testing, showing that VM-DMS can provide a good personalized recommendation service.Finally, the main conclusions are summarized and the prospect of the research is givern.