This research introduces a novel approach to make AI systems better at providing emotional support through optimized conversation strategies.

• Addresses key limitations in current AI support systems: poor strategy selection and preference bias
• Develops a new Chain of Strategy Optimization technique that outperforms traditional supervised fine-tuning
• Achieves measurable improvements in both strategy accuracy and emotional support quality
• Creates more adaptable AI systems that better respond to users' emotional needs

This advancement has significant implications for digital mental health support, potentially expanding access to emotional support resources when human counselors are unavailable.

Chain of Strategy Optimization Makes Large Language Models Better Emotional Supporter