Research on Personalized Recommendation Model of News Content Based on User Behavior Data
Jiachuan Wu, Lingqi Chen · 2023
With the advent of the information explosion era, the massive emergence of news content makes users face the problem of information overload. In order to solve this problem, personalized recommendation technology has become an effective means to improve the user experience. The purpose of this paper is to study the personalized recommendation model of news content based on user behavior data, especially focusing on the combination of deep learning and traditional recommendation algorithms. Firstly, the user behavior data is comprehensively analyzed. On this basis, a deep learning model is introduced, and the user's time-series behavior is modeled through technologies such as recurrent neural network (RNN) to capture the evolution process of user's interests more accurately. At the same time, this paper fully considers the advantages of traditional collaborative filtering (CF) recommendation algorithm to cope with the shortcomings of deep learning model in small sample scenarios. In the model design, we propose an attention mechanism model that combines deep learning and traditional recommendation algorithm. By dynamically adjusting the weight of each part, the model effectively integrates the outputs of the two, fully considers the time series information and multi-source information in user behavior data, and improves the diversity and coverage of recommendations. Experiments on large-scale news data sets show that the fusion model is significantly better than the model that independently applies deep learning or traditional recommendation algorithm in accuracy, diversity and coverage.