Fine-Grained Debate of LLMs for Automated Data Enrichment in Mental Health and Online Safety

This research explores a novel approach to improve data quality for sensitive applications like mental health and online safety using a confidence-aware debate framework. The use of open-source LLMs makes this approach potentially more accessible and cost-effective than proprietary solutions.

• •Applies LLMs to enhance data quality for critical domains.
• •Utilizes a debate-based approach to refine data enrichment.
• •Employs open-source LLMs, increasing accessibility and potentially reducing cost.