Research on Personalized Recommendation System of Short Video Content for College Students Based on Artificial Intelligence

This paper proposes a kind of personalized recommendation system based on AI, which aims at delivering the short video content according to the user's interest and behavior. This system is used to extract feature and recommend model building. Then, it makes use of students' behavior data to optimize the recommendation effect. Firstly, the system preprocesses the user's behavior data, and constructs the user's interest matrix through the analysis of the user's viewing history, likes and comments. Then, the non-negative matrix factorization technique is used to decompose the user's behavior matrix into several latent factors to reveal the user's hidden interest. The system can create a personalized video recommendation list for every user. To enhance the recommendation effect, CNN is used to optimize the feature, to improve the matching degree of the short video content and the user's interest. In the simulation experiment, we compare several recommended algorithms, such as the traditional collaborative filtering, the content-based recommendation and the non-negative matrix factorization (NMF). Experiments show that the NMF recommendation system has good performance in precision, recall and F1. Compared with the traditional method, NMF-based model has a 15% increase in precision, 10% in recall, and 12% in F 1. Especially in university students, the recommendation effect is obviously enhanced, and the user's satisfaction and viewing time are increased.