| With the rapid development of Internet technology,the number of films has grown exponentially.How to effectively select the most satisfying movies in the massive movie network resources has become a obsession for people,so the movie recommendation system came into being.The domestic major video websites have also introduced recommendation modules,but the cold-start of new movies has not been effectively solved.As one of the most popular forms of entertainment,movies have a huge user base.If a newly added movie is not effectively recommended,it will be in a “cold” state for a long time,and it will be a huge loss for users and the movie itself.In this thesis,A method based on Factorization Machines and active learning to solve the cold-start of new movies(the new movie recommendation)is proposed.The main work of the thesis includes the following aspects:1.Aiming at the cold-start problem of zero rated movie,this thesis proposes a new method based on Factorization Machines and active learning to solve the new movie recommendation problem.This method is based on the idea of active learning:First,choose different users for different movies,by formulating user selection criteria,based on factorization and clustering techniques;Second,using the movie ratings of selected users to predict unselected user ratings;Finally,we recommend a new movie to users.2.The experiment uses the MovieLens data set,and at the same time comparing the proposed method with other methods.Experimental results show that the proposed method is superior to other methods on RMSE and MAE.It shows that this method has better prediction accuracy,and at the same time can achieve the recommendation of a new movie to users.3.The method based on Factorization Machines and active learning to solve the new movie recommendation in the thesis can be integrated into the movie recommendation system,and at last designed and implemented a prototype system of new movie recommendation. |
PDF Full Text Request