Research On Product Recommendation Based On Cross-efficiency Evaluation Metho

With the successful development of e-commerce websites and the popularization of rational consumption,consumers are increasingly inclined to collect product information for comparison before purchase.Online reviews have become a potential resource for consumers to obtain real information about products,and research on product recommendation ranking methods based on online reviews has become a hot spot for academic research because of their application value.Scholars have proposed a variety of models to provide consumers with supplementary references for purchase decisions by processing ranking information such as ratings and reviews.After comparing published paper,we have found the limitation of the studies such as neglecting the multiplicative abatement effect of negative evaluation,not considering the time utility of evaluation information,and pursuing satisfaction degree maximization In this thesis,we take the cross-efficiency evaluation method as the basic decision model and come up with innovation of the model and apply it to the commodity recommendation ranking field.Firstly,this thesis addresses the existing deficiencies of the cross-efficiency evaluation model and performs the unique solution constraint and assembly method improvement with the cross-efficiency evaluation model based on the Pareto optimal improvement.The uniqueness of the Pareto-optimal iterative path is ensured by defining the generic initial cross-efficiency value and adding different secondary objectives to the improved model;in addition,the Shannon entropy aggregation method is invoked to embed the aggregation method into each optimization calculation to achieve the fairness of the aggregation of the final cross-efficiency value.Secondly,to increase the scope of application of the improved cross-efficiency evaluation model,the research situation of this thesis is applied by extending the model with multi-period and the presence of undesired outputs.On the basis of the improved cross-efficiency evaluation model,combined with the idea of the multi-period data envelopment analysis model,the total period inputs and outputs are understood as the sum of the inputs and outputs of each period,the efficiency decomposition is carried out through the period input share,and the sub-period efficiency constraint is added to the model to achieve the simultaneous Pareto optimum of the total efficiency and subperiod efficiency;introduction of negative outputs transformation method treatment to achieve positive treatment of undesired outputs through negative treatment of undesired output weights and non-negative constraints on the sum of output weightsFinally,a product recommendation method based on online reviews of websites is proposed.The method considers consumer risk preferences and review time utility,takes into account the purchase cost,applies an extended cross-efficiency evaluation model to give a commodity recommendation ranking method from the perspective of consumer satisfaction efficiency,and uses the ranking of new energy vehicles as an example to verify the validity and feasibility of the method.