| A large number of online platforms conduct purposeful selection of reviews in order to attract more users and increase the reading of the website.However,the problem of redundant information in the selected reviews collection is very serious.Therefore,in order to meet the dual needs of websites and users,it is particularly important to select personalized reviews that are most similar to user preferences.Recent researches have shown that users pay more attention to reviews that are more relevant to themselves when they are browsing and experiencing websites.According to the above problem,this paper proposes a personalized based on user preference selection algorithm,and applies the algorithm in the restaurant reviews.This article has completed the following main work:(1)We process and analyze the historical review through the processing method of text data.The aspects extraction method based on the attention mechanism in deep learning and the hierarchical clustering algorithm are combined to mine users' historical reviews,so as to express users' preferences by considering their own attention.The experiment is carried out through the combination,and user preference is expressed in vector form.(2)We define the review selection problem by taking into account not only the coverage of the review selection(that is,the percentage of the selected reviews in the review collection),but also introduces two selection criteria of the similarity of and the user personalization.On this basis,we propose a personalized selection algorithm and solves the problem of personalized review selection.It sets coverage,similarity and user personalization as the criteria for review selection,and then carries out personalized selection to select a set of K reviews for different users.In order to evaluate the optimization performance of the algorithm,the harmonic means were used to evaluate similarity and user personalization.(3)We apply this algorithm to online restaurant reviews for personalized selection.In this paper,we obtain real and unique user review data from restaurants on the American public review website.The experimental results proved that the method in this article can provide users with personalization based on high review coverage,high similarity,and high user personalization.In this paper,the research on the personalized selection algorithm of online reviews improves the convenience of users' lives. |
PDF Full Text Request