Research and Practice on the Construction of an Agent Paradigm for Vocational Education Based on RAG Technology

Abstract

With the help of artificial intelligence technology, the digital process of education is accelerated, and new requirements for teaching agents in the digital process of vocational education are brought. This study designs a teaching agent paradigm based on RAG (Retrieval-Augmented Generation) technology. The paradigm uses a "data preparation - data retrieval - answer generation"framework to connect and integrate multimodal teaching resources through RAG technology to form a multi-dimensional dynamic knowledge base; uses a hybrid retrieval strategy and re-ranking method to improve the retrieval efficiency of questions; designs role instruction prompt template to normalize the output specifications of large language models and make large language models output specified responses according to different scenarios. This study discusses the construction methods of teaching agents in vocational education and verifies the effectiveness of teaching agents in vocational education through practical implementation, provides technical paths and practical references for the construction of agents in vocational education.