Animation Recommendation System Based On Hybrid Mode

With the development of Internet technology,people get more and more information through computers,and in the process of interacting with the Internet,a large amount of data is generated.If useful information is extracted from these data,the data will further provide humans with better services.The recommendation system analyzes and filters the data generated by the user,and discovers the relationship between data and user,and can also provide much more choices for the user according to different business scenarios.In this paper,a hybrid prediction model is proposed to predict and recommend animation to corresponding users through joint association rules and cluster analysis,which the prediction function implemented in association rules and the description data function in cluster analysis are combined.Firstly,the classification characteristics of each sample set are obtained in the cluster analysis,and then the dimensionality reduction operation is performed by the association rule to better extract the relationship between user and animation,finally making more accurate recommendation.This paper proposes a solution to the cold start problem in collaborative filtering.By classifying users and ranking and analyzing the current hot anime based on heat,the user is given specific recommendations under different categories.The effectiveness of the proposed model is proved by the collaborative filtering algorithm based on traditional association algorithm and the recall rate of collaborative filtering based on clustering algorithm.In order to improve the efficiency of searching and running,this paper uses redis database as the cache of Mysql database.By reducing the disk IO operation of Mysql,the effect of reading user and animation data from the database is further improved.