Research On Recommendation Algorithm Based On Deep Learning And Implicit Feedback

In the context of the era of big data,massive amounts of information have caused many problems for people's daily lives.The recommendation system can effectively alleviate the problem of information overload and help people find information that they are really interested in.Implicit feedback is one of the directions of recommendation system research,which contains a large amount of user behavior data.Research on recommendation systems based on implicit feedback helps to identify potential interests of users and enhance the user's experience.Recently,deep learning has achieved outstanding achievements in the fields of natural language processing and computer vision.Introducing deep learning into the field of recommendation systems and using deep learning techniques to mine potential information in implicit feedback data is expected to improve the performance of the recommendation system.In the research of recommendation algorithms based on deep learning and implicit feedback,the main work done is as follows:1)A deep collaborative filtering model based on attention mechanism is proposed.The model retains the idea of collaborative filtering method.It consists of a generalized matrix factorization model and a deep neural network model based on attention mechanism,which are used to extract linear and nonlinear features in implicit feedback data.The experimental results show that the proposed performance of the model under implicit feedback data is improved compared with the traditional algorithms.2)An implicit feedback recommendation model based on residual network and attention mechanism is proposed.The model is built using a residual network,attention mechanism,and deep neural network to capture the potential patterns in implicit feedback data.A large number of experiments were performed under multiple published data sets and compared to different implicit recommendation algorithms.The results show that the proposed model has improved in different evaluation criteria.