Due to humanity’s close relationships and the quick advancement of AI technology, cultural interaction has become a focus of academic and practical study. Researchers are increasingly focusing on the cultural identity and cultural adaptability of the group of international students in China. In this sense, the paper creates a user portrait model of international students studying in China by utilizing multidimensional labels, making full use of multi-dimensional user data from brief films, and creating an extensive three-dimensional user portrait model. On the basis of the user portrait model, deep reinforcement learning, which can solve the sequential decision problem, is introduced and applied to the recommended videos of Chinese culture dissemination, and corresponding improvements are made by combining the characteristics of the scene. In the performance test of the algorithm, the average probability of users watching the recommended videos of Chinese culture dissemination based on the method of this paper is 0.76, which is higher than that of the comparison algorithm by 0.39, thus proving the superiority of the video recommendation algorithm designed in this paper. This empirical conclusion is derived from the examination of brief Chinese cultural videos created on a specific platform. The frequency of producing unique material for short films has been found to have a strong favorable impact on international students’ behavioral intentions toward Chinese culture in China.
