User Learning Behaviors in Knowledge-Enhanced Large Language Models

Abstract

This study examines differences in user learning interaction behaviors based on knowledge-enhanced large language models (LLMs). It explores the impact of digital literacy and cognitive learning models on user interaction methods. By constructing a systematic framework for analyzing learning interactions, this study integrates in-depth interviews, experimental research, and statistical analysis to reveal significant behavioral variations in knowledge acquisition, problem-solving, and information reconstruction. Experimental results indicate that users with low digital literacy tend to increase query frequency and adjust search strategies to compensate for limitations in information processing. In contrast, users with higher digital literacy demonstrate stronger expression adjustments and enhanced knowledge integration. Additionally, cognitive learning models significantly shape interaction patterns. Evaluative learners engage in more in-depth conversations and intensive reading, while receptive learners rely on the model's direct outputs and rarely reconstruct information. The findings deepen the understanding of user learning behaviors in knowledge-enhanced LLMs. They also provide a theoretical foundation for optimizing personalized learning support and intelligent interaction design. Future research will explore more complex interaction factors and incorporate personalized user characteristics to enhance the effectiveness of large language models in intelligent learning environments.