Sarah Huang, Yinggang Sun, Xiangzhan Yu
This paper presents SP-TeachLLM, a novel framework that leverages large language models (LLMs) to deliver intelligent tutoring for computer science education. SP-TeachLLM integrates advanced AI techniques with established educational theories to enable personalized and adaptive learning experiences. Its core innovation lies in a multi-module collaborative architecture that encompasses curriculum decomposition, multi-strategy generation, reflective learning, and memory augmentation. Comprehensive experiments are conducted to evaluate the system’s effectiveness in enhancing knowledge mastery, problem-solving ability, and teaching performance. The results demonstrate that SP-TeachLLM significantly outperforms conventional approaches, providing valuable insights into the application of AI in education and advancing the development of next-generation intelligent tutoring systems.