Research And Application Of Personalized Movie Recommendation Algorithm Based On Temporal Context

In the era of big data,with the rapid development of the film and television industry,various film and television resources have emerged.With the continuous accumulation of movie resources,the problem of "information overload" is becoming more and more serious in the film and television industry.The application of the recommendation system can effectively alleviate the problem of movie overload.Some video websites use traditional collaborative filtering recommendation algorithms to recommend movies to users,but the algorithm does not consider the time context sufficiently and the recommendation accuracy is not high.In real life,people’s interest preferences will change with time offset,so it is of great significance to study the impact of time context on the recommendation algorithm.In view of the impact of time context information on the recommendation algorithm,this paper mines the user’s historical rating data and uses the user rating time to model the user.The main work of the paper is as follows:Firstly,considering the law of human memory forgetting,a time decay function is obtained by fitting the Ebbinghaus forgetting curve,which can track the user’s interest shift.Considering that the rate of forgetting things at different ages is different,the fitting time decay function is further improved to track the interest offset of users of different ages.And the memory forgetting time function is integrated into the user similarity calculation to realize the score prediction based on the user group.In addition to the law of forgetting,there is the possibility of recalling similar items when people see an item.This phenomenon is called memory activation.Therefore,the time activation function is obtained by referring to the theory of memory activation.This time function can measure the weight of the historical behavior items of the user to the current item,and further obtain the score prediction based on the item group.Secondly,considering the influence of user group and goods group on users,the mixed score prediction is realized.In this paper,this hybrid scoring prediction method is called HMC-CF(Collaborative Filtering based on Human Memory Characteristics)algorithm,and it is verified by experiments on Movie Lens data,which proves the superiority of HMC-CF algorithm.Finally,in order to further verify the feasibility of the recommendation algorithm proposed in this paper,a personalized movie recommendation system is designed and implemented by drawing on the operation mode of current popular video website.The system has movie classification navigation,precise search,popular movies,latest movies and personalized movie recommendation functions.The personalized recommendation module can track the user’s interest deviation,and better meet the user’s personalized and diverse viewing needs.And the interface is neat and beautiful,which can improve the user experience.In this thesis,there are 37 figures,15 tables and 76 references.