Research on a Three-Level Integrated AI Teaching System Driven by Dynamic Knowledge Graphs

Abstract

Addressing common challenges in higher education such as complex course knowledge points, fragmented student cognition, and weak practical abilities, this study constructs a three-level integrated AI teaching system characterized by "global navigation via teaching knowledge graphs, in-depth empowerment through AI toolchains, and dynamic iteration of personal knowledge graphs." Teaching experiments were conducted in pilot institutions, where three core graphs—target graphs, dynamic knowledge graphs, and networked problem graphs—were established at the teaching graph level. At the AI tool level, intelligent data analysis was realized through automatic code generation, modeling, and simulation. At the personal graph level, the DeepSeek model was employed to generate personalized knowledge mind maps, which in turn fed back into the teaching knowledge graphs. Experimental results demonstrate that this system effectively enhances students' knowledge integration and practical capabilities, providing a scalable paradigm for the digital transformation of higher education.