Study On Movie Recommendation Algorithm Based On Deep Learning

In today’s era,major mobile Internet companies are fiercely competitive,such as iQiyi,Youku,and Tencent in the film and television industry.In order to ensure their core competitiveness in the industry,enhance user stickiness,and thereby make a profit,a good movie recommendation algorithm is essential.In addition,excellent recommendation algorithms can also facilitate the public,improve the efficiency of the tertiary industry.At present,the most widely used recommendation algorithm is the traditional collaborative filtering algorithm,but as the amount of data increases,the disadvantages of the traditional recommendation algorithm gradually become apparent,the particularly serious performances is the problem of sparse data and low information utilization,so in some occasion,the recommendation effect is not very good.To alleviate this problem,this paper proposes a movie recommendation model based on deep learning technology.The data set used in this article is the movieslens data set.First,the exploratory analysis of the data is used to find the distribution and general rules of the data.The results show that the data conforms to the long tail distribution.Second,the deep learning recommendation model framework is constructed.The two parts in parallel are the user feature extraction network and the movie feature extraction network,and the third part is the mean square error of the prediction score and the true score.The entire model mainly uses fully connected networks.In the movie feature extraction network,text convolutional neural network is used to extract movie title features.Then,training the network and using the adam algorithm to continuously iterate in order to minimize the mean square error of the loss function.Finally,combining the user and movie feature to predict scores,thereby generating a recommendation list to complete the recommendation task for the target user.In order to confirm the superiority and effectiveness of the deep learning recommendation model in this paper,a comparative analysis of the deep learning recommendation algorithm and the traditional collaborative filtering movie recommendation algorithm is also made in this paper.The evaluation indexes of collaborative filtering algorithms based on users,items,and matrix decomposition are calculated.Compared with the evaluation indexes of recommendation algorithm based on deep learning,the final evaluation results are obtained,the deep learning recommendation algorithm is indeed superior to the traditional collaborative filtering recommendation algorithm,and its superiority is specifically reflected in its ability to effectively alleviate data sparsity and using deep networks to get deeper features.This gives us a way toimprove the accuracy of recommendations,such as quoting more external related information or using more effective models that can extract deep features.