Practical application of large language models in educational informatics: specification, framework, and applications

Abstract

To address the issues of complex implementation, deep model coupling, and poor reusability in deploying large language models (LLMs) in higher education, a set of application standards for LLMs in educational informatics was proposed. The standards covered API interface designs from three perspectives: general conversation, retrieval enhancement, and intelligent agents. Based on these standards, a technical implementation framework was provided. The paper demonstrated how large language models could be deeply integrated into campus life to improve teaching quality and learning experiences through three specific cases at East China Normal University (ECNU). The results show that applications developed based on a standardized framework possess good interactivity and extensibility, and the efficient reuse of underlying capabilities significantly improves development efficiency. Finally, the paper summarized the application scenarios of LLMs at ECNU and looked forward to future trends, emphasizing the importance of integrating with educational contexts. This provided a reference for promoting the widespread application of artificial intelligence technology in the field of education.

Keywords

• large language model
• educational informatic
• specification
• framework
• applications
• AI