Construction of an AI-driven personalized training system for live streaming scripts and verification of educational effects
Abstract
We propose an AI-driven personalized script training system for live streaming, integrating natural language processing (NLP) and reinforcement learning (RL) to enhance educational effectiveness. The system automatically generates base scripts using an NLP module and then personalizes them through an RL strategy that adapts to individual user performance. A reward function is designed to capture key metrics such as audience engagement and learning outcomes, guiding the RL agent in optimizing script delivery. The overall architecture operates in a closed loop: script suggestions are generated, tried in practice sessions, and then refined based on feedback. Experimental validation demonstrates that this approach improves presenter engagement and audience learning outcomes, highlighting the potential of AI-driven personalization in educational live streaming.
Authors

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.