Tag
This paper introduces DiPS, a Q-learning framework that dynamically selects persuasion strategies for high-stakes scenarios like wildfire evacuations, achieving higher success rates than zero-shot LLM and RAG baselines.