Research on online psychological teaching and behavior analysis of college students based on improved collaborative filtering

Given the importance of mental health to the quality of human life and social development, as well as its central position in education in the new era, this paper focuses on the critical group of college students, aiming to optimize the teaching mode of online mental health education through technological innovation, to improve their psychological quality and coping ability. Given the significant changes in college students’ lifestyles, learning modes and communication styles in the Internet era, this paper profoundly analyses these changes’ positive and negative impacts on mental health. In particular, it is pointed out that the incidence of mental health problems among Chinese college students has risen to more than 20%, with depression, anxiety, Internet addiction, and emotional problems being particularly prominent, which seriously constrain students’ academic achievements and quality of life, and may even induce undesirable behaviors. In order to effectively deal with this situation, this paper identifies several key factors affecting the effectiveness of learning based on an in-depth analysis of students’ online psychological learning behaviors. It proposes an optimized collaborative filtering algorithm accordingly. In order to overcome the performance bottleneck and sparsity problems that traditional collaborative filtering algorithms may encounter when dealing with large-scale datasets, we propose an improved collaborative filtering algorithm based on a hybrid strategy, which achieves fine-grained modeling of students’ psychological and behavioral patterns by combining the advantages of SVD++ matrix decomposition, neural network embedding, and content filtering, and through the fusion of multi-source data and feature intersection. The algorithm significantly improves the prediction accuracy of students’ learning behaviors and achieves an increase of about 30% in the accuracy rate compared with traditional methods. This research result provides a powerful tool for accurately identifying students’ psychological needs and potential problems. It lays a solid foundation for universities to carry out personalized and high-performance online mental health education models.