Research Of Recommendation Algorithm Based On Convolutional Neural Network

With the rapid development of the Internet,when faced with massive amounts of information on the Internet,users cannot rely on traditional search engines to effectively filter out the content they need,and the recommendation system technology came into being.Collaborative filtering recommendation algorithm is widely used in information filtering and recommendation system due to its good scalability and ease of use.However,the accuracy of collaborative filtering algorithm will greatly reduce due to the sparseness of user's rating information.At the same time,when the user's interest changed,the traditional collaborative filtering algorithm cannot correct the recommendation results in time,and the real-time performance is poor.In recent years,neural networks have been widely used in various fields due to their advanced performance and high-quality feature extraction capabilities,which has also brought new development points for recommendation systems.First,for the sparseness of traditional collaborative filtering algorithms,based on the inherent advantages of CNN high-quality feature extraction in convolutional neural networks,this paper proposes a CNN-CF algorithm based on convolutional neural networks.CNN is used to learn user preferences and find the potential relationship between user and items,so as to fill the user rating matrix with high quality,alleviate the sparsity problem,and conduct experimental comparison of the algorithm.Although the CNN-CF algorithm improves the recommendation accuracy of the traditional collaborative filtering algorithm,it is still a recommendation based on the user's historical behavior information.The recommendation results of the CNN-CF algorithm reflect the user's historical average preference.It is suitable for offline recommendation services with stable interests.Aiming at online recommendation services with large changes in interest,time weights are introduced to modify the recommendation results of the CNN-CF algorithm,and the CNNT-CF algorithm is proposed and verified through experiments.Finally,this paper designs a recommendation framework based on CNN recommendation algorithm,including: user interaction module,data storage module,offline recommendation module,and online recommendation module.The CNN-CF algorithm is applied in the offline recommendation module to implement offline recommendation.The CNNT-CF algorithm is used in the online recommendation module and is responsible for online recommendation.