Research On Product Ranking Model Based On Online Reviews

With the development of information technology and human society,e-commerce platforms are becoming more and more popular.Meanwhile,large E-commerce platforms such as Tmall.com,JD.com and amazon have launched a review system for products and services on their platforms,which enables consumers to share their real purchasing experience or obtain information related to products or services from online reviews.Meanwhile,people are encouraged by e-commerce platforms to use texts,pictures or videos to describe what they have bought.Also,potential consumers are able to read these online reviews and thus decide whether or not to buy the products.However,due to the huge volume of online reviews,it is difficult for consumers to read all the reviews and extract important information in a short time in order to select the desired products.Therefore,the ranking products based on online reviews information has become a hot spot in the field of management and decision making.However,most of the existing research on ranking products with online reviews information is based on the theory of directed graph or weighted directed graph,which only considers the positive and negative emotions of online reviews information.The current studies do not involve the emotional intensity of online reivew information,which will lead to information loss.In addition,the previous studies focused on the products,which did not consider the attributes for products and cannot effectively merchandise personalized recommendation for consumers.Further,existing studies used TOPSIS method to assess and rank online products.However,this method is based on the assumption that decision makers are completely rational.Therefore,the ranking results based on this method is not rational.Recently,more and more researchers focused on TODIM method,which considers the loss aversion of decision makers in the process of constructing dominance function.However,existing studies on TODIM method can not deal with all attriture interactions.Therefore,in this study,we extend TODIM method to rank products with online review information.Specifically,from the perspective of measuring the sentiment of product features,this research established the sentiment measurement model for product features under intuitionistic fuzzy environment and hesitant fuzzy environment,respectively.In addition,from the perspective of optimizing multi-attribute model,this research proposed the 2 additive Choquet intuitionistic fuzzy TODIM method and hesitant fuzzy TODIM method,where the attribute interaction index is used to describe all the interactions among attributes.The proposed model can be applied to handle decision making problems with attribute interactions.In the following parts,this research concludes the main focuses and findings:(1)Constructing an intuitionistic fuzzy emotion measurement model and a hesitant fuzzy emotion measurement model based on online reviews information.Consideraing the emotion intensity of product features and limitation of numerical values to describe decision-makers' subjective judgements,this research construct the fuzzy emotion measurement model to obtain the emotion intensity of product features.Based on the<feature word,emotion word,emotion modifier>,the proposed model can calculate the sentiment score of product features.Also,the empirical examples also indicate that the proposed model can describe the emotion intensity of consumers for different product features,which provide efficient data for the following proposed product ranking models.(2)proposing the improved intuitionistic fuzzy score function and the hesitant fuzzy score,respectively.First of all,this reseaerch studies the existing literature about the deficiency of intuitionistic fuzzy scoring functions.Then,by considering the hesitation degree distribution and the thought of voting model,this research thus proposed the improved intuitionistic fuzzy score function.In order to better ranking the intuitionistic fuzzy numbers,this paper further puts forward the intuitionistic fuzzy accurate function and define the ranking rules of intuitionistic fuzzy number.This studies leveraged the example of numerical experiments to prove the rationality of the intuitionistic fuzzy scoring function.Secondly,in view of the irrationality of the existing hesitating fuzzy score functions,this research considers the complement of elements in the hesitating fuzzy numbers and the simplicity of the calculation steps,and thus proposed an improved hesitant fuzzy score function.Accordingly,the sorting rules of hesitating fuzzy Numbers are defined.This function simplifies the calculation procedure and improves the ordering efficiency of the hesitant fuzzy Numbers.(3)constructing the 2 additive Choquet-based intuitionistic fuzzy TODIM method and 2 additive Choquet-based hesitate fuzzy TODIM method,respectively.Considering that the existing TODIM multi-attribute decision model cannot well deal with the relations of redundant association,independent association and complementary association among attributes,this research proposed 2 additive Choquet-based intuitionistic fuzzy TODIM method and 2 additive Choquet-based hesitate fuzzy TODIM method based on 2-addiitive fuzzy measure and choquet integral,respectively.The models leverage the attribute interactive index to describe different relations between different attributes.Finally,the results of comparative analysis and sensitivity analysis show that the risk attitudes of decision makers have influences on the ranking results.Finally,the empricial results show that the proposed models can rank products efficiently,which can be applied in E-commerce platforms.