Intuitionistic Fuzzy Sets Construction And Merchants Ranking Based On Text Mining

With the continuous upgrading of the Internet consumption model,consumers have more choices and more information to face.Online consumption decision-making has gradually become a daily problem faced by people,and O2 O merchants selection is one of them.Intuitionistic fuzzy sets have advantages in representing and processing fuzzy information,and the determination methods of membership degree,non-membership degree and hesitation degree of intuitionistic fuzzy sets directly affect relevant applications.In order to better solve the problems of overloaded information and different preferences in Internet O2 O merchants selection,we use actual data to determine the IFS,improve the ability of text mining,and propose a ranking method of the alternative merchants based on the previous two methods.This has important practical value for both consumers and merchants in the context of big data.The thesis proposed the membership determination method of intuitionistic fuzzy sets based on deep learning and the aspect level sentiment classification model.On this basis,the ranking method of O2 O alternative merchants based on online reviews was proposed.The main contents are as follows:First of all,in order to solve the problem of insufficient practical application of existing methods,a deep learning-based method was proposed to determine the membership degree,non-membership degree and hesitation degree of IFS using actual data.The experiment was carried out in the text data and good results were obtained.This method breaks through the technical limitations of traditional methods and expands the application range of IFS.Secondly,mining the aspect level sentiment orientation of online reviews is an important component of the Internet consumer decision-making problem.Therefore,a dual attention BILSTM based aspect level sentiment classification model was constructed,which applied selfattention and aspect-context attention to better identify the aspect sentiment in reviews.The experiments results showed that the model performed better than the traditional model and the single model,and the outputs of the model could be further applied to the Internet consumer decision-making problem.Finally,in the problem of the ranking of alternative merchants based on online reviews,the outputs and the evaluation index of the sentiment classification model were applied to determine the IFS elements,which reflected the support,opposition and abstention towards consuming decisions.In this process,both the hesitation of neutral sentiment and the uncertainty caused by limited accuracy of the model were considered.Then the intuitionistic fuzzy TOPSIS method was applied to rank the alternative merchants.In the experiment,six restaurants on the websites were ranked.The results were compared with the score ranking and the sales ranking respectively,which showed the effectiveness of this method.Besides,suggestions of different aspects were provided for merchants to improve management and operation.The application shows that combining the IFS with aspect level sentiment classification to solve the Internet alternative merchants ranking has theoretical and practical significance.The proposed methods extend the determination method of IFS,construct the dual attention BILSTM model for aspect level sentiment classification,and provide a new solution to the Internet consumer decision-making.