Research On Recommendation Method And Its Application Based On Implicit Feedback

In the past few years,with the vigorous expansion of the Internet,web information has grown geometrically and has gradually developed into an information ocean,but the use of information is inefficient.In order to reduce the occurrence of this problem,a personalized recommendation system has been created,which plays an increasingly important role in information filtering and user personalized recommendation.As a key part of the personalized recommendation system,the user interest model plays a decisive role.The construction quality of user interest model in a recommendation system directly determines the recommendation accuracy and user satisfaction of the system.Therefore,the research on the interest model has important application value and significance.At present,domestic and foreign scholars have conducted a large number of related research work on interest modeling and updating issues,and have achieved fruitful research results.However,in the study of user interest modeling and updating,the following problems still need to be solved: First,when building a user interest model,they often only focus on the user's browsing content or related actions when the user browses,and lack of both.Combined with the study of interest vectors,secondly,after establishing an interest model,many studies simply divided and judged user interest changes with the time curve,and did not reflect the continuous change of user interest on the interest model to analyze user interest.This paper takes the establishment of an efficient interest model as the research goal.It mainly focuses on the measurement of user interest and the update of interest model.It also establishes a recommended prototype system with practical application value.The contributions and innovations of this article mainly include the following aspects:(1)In view of the lack of consideration of user page browsing rate in the existing user interest model,and the lack of integration of user interest into the construction of the interest model,a user interest model construction method based on implicit feedback is proposed.Improve the accuracy of the recommendation.This paper proposes a method of constructing user interest model based on implicit feedback.First of all,it analyzes the user interest from user browsing behavior,and establishes the user's classic implicit feedback behavior and interest in page browsing time,browsing speed,save page,and favorite page.The relationship between degrees;then,based on the implicit feedback behavior,the concept of topic interest is proposed,and it is combined with the traditional interest model based on VSM to construct user interest based on topic interest.By comparing with the traditional VSM model method on the data set collected by professional software,the experimental results show that this method obviously improves the accuracy of model recommendation.(2)For the problem that the existing user interest model cannot be updated in time according to the user's own interest offset,an implicit feedback recommendation method based on the user interest update model is proposed,which improves the model recommendation accuracy rate.This paper proposes an implicit feedback recommendation method based on the user interest update model.First,the content of each web page browsed by the user and the degree of interest of the web page calculated by the method of measurement of mixed interest degree are converted into a vector;then,vector aggregation is performed.The interest model is updated after the operation.Compared with the recommendation method based on K-means clustering model on the professional software acquisition data set,the experimental results show that this model improves the recommendation accuracy rate.(3)Based on the above research results,a recommendation prototype system based on user interest model is designed and implemented to provide personalized recommendation services for users.