New media user behavior prediction and content recommendation based on deep learning
L. Xia · IET conference proceedings. · 2023
With the development of information age, mobile apps have penetrated into every aspect of modern people's life, accompanied by the explosion of user information, video information and product information. How people get the information they are interested in, and how apps and web pages can push users more interesting content from massive data based on their browsing history and other relevant user characteristics have become popular research nowadays. In recent years, user click prediction models and recommendation models designed by combining deep neural networks have been far more effective than traditional machine learning models. With the explosive growth of data sets, effective processing of data information combined with neural network models has become the key to improve the effectiveness of the models. Therefore, in this paper, we improve and optimize the user behavior click-throughrates (CTR) prediction algorithm on apps/web pages by starting from the user click-throughrates (CTR) model in sparse scenarios. Finally, it is proved experimentally that verifying the feature interaction information is the key to the user click-through prediction model, which provides new ideas and methods for the design of the user click-through prediction model.