Analysis And Implements Of Hybrid Recommendation System Based On Spark

In the era of big data,recommendation systems have become an important means to solve information overload.However,the recommendation system is still facing many problems,such as data sparsity,cold start,and timeliness.To solve the above problems,this paper proposes a hybrid algorithm of joint content recommendation,Alternating Least Squares(ALS)recommendation,and Neural Matrix Factorization model(Neu MF)recommendation.Then,a hybrid recommendation system in the field of movies was implemented on the Spark platform,which can improve the accuracy and timeliness of recommendations to a certain extent.The main contents of this article are as follows:First,for the data sparsity and cold start of the traditional recommendation algorithm,a hybrid recommendation method of fusion content recommendation and ALS recommendation is used.On this basis,a Neu MF algorithm is proposed to solve the lack of linear expression ability and content of the ALS model.Secondly,on the proposed hybrid recommendation algorithm,a hybrid recommendation engine and Web application are designed,which combines Pyspark and Tensor Flow to simultaneously train content recommendation,ALS recommendation and Neu MF model on a distributed platform,and uses Python to implement a movie hybrid recommendation system.And has the user-friendliness of the software interface.Again,the comparison of the operating efficiency under different numbers of nodes and data volume shows the superiority of Spark in big data processing.By controlling the variables,the parameters of the mixed model are optimized,and the performance of different algorithms of the mixed model is compared.The results show that the hybrid recommendation algorithm proposed in this paper greatly improves the hit rate.In this paper,the traditional recommendation algorithm and Neu MF neural network algorithm are combined on Spark.The experimental results show that the former can be better combined with the expressive advantages of the latter on a distributed platform.Based on the proposed The movie recommendation system implemented by the hybrid recommendation algorithm can provide users with more personalized and smoother information services.