The study of course recommendation system is critical in students’ social relations, however it has an issue with erroneous performance positioning. The typical Particle swarm arithmetic is unable to address the inaccurate recommendation positioning issue in students’ social relations, and the result is insufficient. As a result, a Neural network model-based research on course recommendation system is provided, and the research on course recommendation system is assessed. To begin, the alternating neural network theory is used to discover the influencing elements, and the indicators are split based on the study of course recommendation system’s needs to decrease interference factors in the study of course recommendation system. The alternating neural network theory is then used to create a Neural network model study of course recommendation system scheme, and the outcomes of the study of course recommendation system are thoroughly examined. The MATLAB simulation results reveal that, under particular evaluation conditions, the Neural network model outperforms the standard Particle swarm arithmetic in terms of study of course recommendation system accuracy and time of influencing variables.