Design And Implementation Of Movie Recommendation System Based On Sequential Recommendation Algorithm

With the rapid development of the Internet,the era of information explosion coming,and the problem of information overload is becoming increasingly serious.As an important tool to alleviate information overload,recommendation system can fully understand users and items,recommend the most suitable items for users,and reduce the time and cost of information processing.Although there are many software that can provide movie recommendation,due to commercial considerations,their recommendation mainly serves its business objectives,and requires users to fully accept all kinds of other services besides recommendation,which is too redundant.In addition,the traditional recommendation algorithm often regards the user’s historical interaction records as a collection,with inadequate mining of the sequential relationship between items and insufficient understanding of user’s interests.Based on the above problems,this paper delves into the sequential recommendation algorithm and designs and implements a movie recommendation system combined with the proposed new algorithm,with user experience as the primary goal,light enough to focus only on providing users with movie recommendations that match their interest preferences.The main work and innovation of this paper are as follows:(1)An attention network model based on long-and short-term interests,LSISA,is proposed,which uses attention networks to jointly model users’ longand short-term interests,and models the evolution of users’ long-term interests using serial modeling.Experimental results on four real-world public datasets show that LSISA effectively improves the accuracy of recommendations.(2)The recommendation algorithm alone is difficult to implement directly in complex scenarios in industry,so the paper designs and implements a complete recommendation chain to ensure the stability of the recommendation results.The whole recommendation process is divided into two stages:recall and rank,to filter the candidates set.LSISA is used as the main recall algorithm,combined with rule recall and general vectorization recall algorithm to form a multi-channel parallel recall,and DIEN algorithm is used as the main rank algorithm to rank the items.(3)We designed and implemented a neutral and lightweight movie recommendation system,Zhiying Movie Recommendation System.According to the system development process of software engineering,the system was realized through requirements analysis,outline design,functional module division and detailed design of the system,and system development and testing.The frontend of the system is based on WeChat Mini Program and Vue respectively,and the back-end is based on SpringBoot to realize the separation of front and back ends.The algorithm module uses microservices to provide services.The database uses Mysql and Redis to store data.The system developed a core recommendation module to provide movie recommendation,and developed three basic function modules:user module,movie information module and management module to ensure normal system use.The final system test results show that the system meets the initial expectations and can satisfy users’ needs by providing them with movie recommendations that match their interests,Based on the above research contents and work,this paper designs and implements a movie recommendation system,Zhiying Movie Recommendation System.The final functional and performance tests of the system show that the system focuses on movie recommendation function,has good recommendation accuracy,and is directly based on WeChat Mini Program,which reduces the user’s usage burden,realizes light weight and has good practical value.