When using the current method to intelligently recommend short videos,the user's long-and short-term preferences are not considered separately,which leads to the problems of low click-through rate and recommendation accuracy,as well as the poor comprehensive recommendation effects.Therefore,a short video intelligent recommendation algorithm technology based on deep learning is proposed.Firstly,the scene features and behavior features of the target short video are extracted to obtain joint features.Then,deep learning are used to model and analyze the user's long and short-term preferences.Finally the similar-ity and preference are calculated by joint features of the target short video and the the joint features of the user's short video preference,and several short videos with the highest preference are selected to form a recommendation list to complete the intelligent recommendation of short videos.The experiment results show that the proposed method has higher click-through rate,recommendation accuracy rate and better compre-hensive recommendation effect.
关键词
深度学习/短视频/智能推荐/帧间匹配法/图卷积神经网络
Key words
deep learning/short video/intelligent recommendation/inter-frame matching method/graph convolutional neural network
维普
万方数据