“Youth night school” is a mode of education and training that provides vocational and academic training during free time at night, and is an important way for young people in society to improve themselves. This paper discusses in detail the participating members and corresponding responsibilities in the “Youth Night School” in higher vocational colleges and universities, as well as its organizational framework model. The hypergraph neural network algorithm is used to capture complex relationships in educational data based on the existing “Youth Night School”. The mapping function is constructed by multilayer perceptron (MLP) to eliminate the information bias between the domains, so as to realize the information conversion between the domains. We also design a rating prediction and construct a course recommendation system for “Youth Night School”. Comprehensive above, the formation of higher vocational colleges and universities “youth night school” education model, the design of comparative experiments, application experiments to verify its effectiveness. The recommendation algorithm designed in this paper has a hit rate of up to 0.874 and a coverage rate of more than 95%, which can accurately and comprehensively recommend the courses of “Youth Night School” and improve the teaching effect.
