Intelligent TV Recommendation Algorithm Based On User Behavior Analysis

With the rapid development of information technology,the generation of massive data has brought about information overload and selection barriers.Each user has limited time and energy.Faced with massive information,how to help users filter and filter data in a short time.It is of great practical significance to select and obtain valuable information.In the face of "information overload",according to the historical behavior data accessed by users in the system,deep interest point mining is carried out to provide users with personalized and accurate information recommendation,which not only meets the essential information demands of users,but also maximizes the enterprise's own Therefore,personalized and accurate information recommendation contains unlimited business opportunities.This paper takes the smart TV system as the background,according to the user's usage habits and the access data in the system,and uses the advanced recommendation algorithm technology to study the user characteristic behavior recommendation algorithm model based on the smart TV system.In order to improve the accuracy of the recommendation algorithm,the traditional recommendation algorithm is improved to form a hybrid recommendation algorithm based on the improved Slope One.Very good effect,suitable for smart TV terminals,worthy of promotion and application in the field of smart TV The main contents of this paper are as follows:First,the research background of the smart TV recommendation algorithm and the significance of the current situation in practical applications are summarized,and the domestic and foreign progress and future development trends in the field of smart TV recommendation algorithm technology are studied;the working principle of the smart TV recommendation system is studied.,analyzes the recommendation algorithm model and implementation process.Second,preprocess the original data set of user behavior,improve on the basis of the implicit scoring model based on user viewing duration and set video scalar,and introduce factors such as frequency of on-demand and program viewing duration to construct a new model based on user viewing.The implicit scoring model of behavior is analyzed experimentally on the preprocessed dataset to verify its rationality and applicability.Third,combined with the characteristics of smart TV data information,the user collaborative filtering algorithm based on neighbors is selected,and the common scoring item weighting factor is introduced to optimize and improve the similarity calculation method.In order to alleviate the data sparsity,combined with the popular learning spectrum clustering method,the original data set was clustered and analyzed,and finally a collaborative filtering algorithm(MLCF+)based on similarity optimization and popular learning was formed.And through a series of comparative experiments,it is proved that the recommendation accuracy of the MLCF+algorithm is higher.Fourth,consider only referring to the implicit score and ignore the deviation caused by user characteristics and item similarity.Based on the Slope One algorithm,item similarity and user characteristic factors are introduced for improvement and optimization,forming a method based on item similarity and user characteristics.The Slope One algorithm.Combined with the traditional prediction and scoring method,a smart TV recommendation algorithm based on user behavior analysis is finally formed,which provides a new solution for the personalized and accurate recommendation of smart TV system programs.Through comparative experiments,it has better performance on the two evaluation indicators of MAE and RMSE,that is,it has better recommendation accuracy.