This research develops a hybrid approach to accurately identify the types of social support patients seek in online health communities, addressing the challenge of limited labeled data.

• Combines semi-supervised learning with LLM-based data augmentation to improve classification performance
• Strategically uses LLMs to generate balanced training examples across support categories
• Implements a two-stage framework that significantly outperforms traditional machine learning approaches

For healthcare platforms, this technology enables more accurate matching between patient questions and appropriate support responses, potentially improving patient satisfaction and health outcomes.

Understanding Social Support Needs in Questions: A Hybrid Approach Integrating Semi-Supervised Learning and LLM-based Data Augmentation