This paper analyzes the technology of university English digital teaching formation in the context of the Internet, designs a university English resource base according to the characteristics of digital teaching resources, the degree of integration, and designs an autonomous teaching platform for university English based on the English resource base. On this basis, combined with the matrix decomposition algorithm, the knowledge point association matrix integrating multiple factors was constructed to design the English resource recommendation algorithm based on the matrix decomposition of the improved hidden semantic model. In addition, the Jmeter testing tool and SPSS tool were utilized to analyze the performance of the system and the application effect of the teaching model. The results show that the overall recommendation accuracy is also above 0.92, the request anomaly rate is 0% when the number of threads is 300, the response times are all kept within 0.1s, and the overall English learning effect is improved by about 0.3 on average. The digital teaching mode of university English in the context of the Internet can greatly enhance the impact of English teaching.