Statistical Recommendation Method And Application Based On Double Hierarchy Linguistic Term Set

In the age of data with explosive growth of information,it is difficult for users to quickly get the information they really need from massive data.The recommendation algorithm uses historical rating data to show the information or goods users are most likely to be interested in.In practical application scenarios,people tend to use natural language in description and evaluation,and the information is accompanied by a certain ambiguity,but the existing recommendation algorithms have not made full use of language information.In this paper,the fuzzy tool of double hierarchy linguistic term set(DHLTS)is innovatively introduced into statistical recommendation method.And we discuss the collaborative filtering method based on DHLTS and its application.Deep learning recommendation method based on DHLTS is constructed and personalized recommendation is realized.On this basis,we introduce the double hierarchy linguistic ordered weighted logarithmic averaging distance(DHLOWLAD)into the deep learning recommendation method.The main contents are as follows:(1)Collaborative filtering recommendation method and application based on DHLTS.This paper establishes the basic rules of transforming user comment text into DHLTS,and the collaborative filtering recommendation model based on DHLTS is constructed.According to the user review data set provided by Amazon e-commerce shopping platform,it can predict the rating of target users on random goods and make personalized recommendation.(2)Recommendation method and application of deep learning based on DHLTS.Due to the head effect and weak generalization ability of collaborative filtering,a deep learning recommendation model based on DHLTS was established.After the review text is converted into DHLTS form,it is converted into one-dimensional form suitable for model calculation,and the loss function is redesigned according to the distance formula of DHLTS.Finally,the most likely interested movies and similar movies are recommended for users.(3)Statistical recommendation method of deep learning based on double hierarchy linguistic distance measure operator and its application.Because of the simple double linguistic term set distance formula is easy affected by extreme deviation value,the double hierarchy linguistic term set(DHLTS)in ordered weighted logarithmic averaging distance(OWLAD)operator of the promotion,get double hierarchy linguistic ordered weighted logarithmic averaging distance(DHLOWLAD)operator,in deep learning recommended approach to further improve the loss function,DHLOWLAD is used instead of cosine similarity to find items with similar features.Finally,the model is applied to movie recommendation to verify the accuracy of the model.To sum up,DHLTS is systematically introduced into the recommendation model in this paper,and the natural language information in user evaluation is used to remedy the problem of insufficient recommendation utilization information.However,it also has disadvantages:when the number of users is far greater than the number of items,the user similarity matrix will be very expensive.In the future,the double hierarchy hesitant fuzzy linguistic term set can be incorporated into the model without considering the hesitation of users between different evaluations.