Research And Implementation Of Personalized Recommendation Algorithm Based On Time Effect

Recommendation system is a product that emerged in the era of big data and intelligent information.It has been successfully applied to various scenarios such as commodity recommendation,food recommendation,travel recommendation,video recommendation,social recommendation and so on.At present,most recommendation systems pay more attention to explicit feedback.However,in practice,implicit feedback is much richer than explicit feedback,and even in some scenarios only implicit feedback can be obtained.Therefore,the importance of implicit feedback recommendation is self-evident.At the same time,the traditional recommendation algorithms based on implicit feedback mostly ignore the influence of time factors on the recommendation results.In view of the above problems,this paper improves the recommendation algorithm in two scenarios and designs a recommendation system for children's video.The specific work of this paper is as follows:(1)Aiming at the problem that the traditional time weighting algorithm does not mine the user's forgetting rule of "first fast then slow",this paper designs a time window function,improves the traditional time weighting strategy,and integrates the improved time weighting strategy into the fast matrix factorization algorithm.Finally,the comparison experiments are carried out on two public datasets to verify the feasibility and effectiveness of the improved algorithm.(2)In view of the fact that the traditional GRU model neglects the time when users click on commodities in time series data,this paper adds a time gate to the traditional GRU model,proposes a Time-GRU model,and explores the impact of two loss functions on the Time-GRU model.Experiments show that the GRU model with time gates can improve the performance of recommendation tasks.(3)This paper designs a recommendation system for children's video based on B/S architecture,and integrates the above improved algorithm into the algorithm module of the system,constructs the overall framework of the system and the software logic system,establishes the database table,and finally completes the development,testing and deployment.