This research introduces a collaborative stance detection approach that combines Small Language Models (SLMs) and Large Language Models (LLMs) to efficiently analyze social media attitudes.

• Creates a more resource-efficient system for real-time social media monitoring
• Reduces dependency on computationally expensive LLMs without sacrificing accuracy
• Implements consistency verification between SLM and LLM predictions
• Enables practical deployment for security applications processing vast amounts of social data

For security professionals, this approach offers a scalable method to detect potentially harmful stances and misinformation campaigns while maintaining operational efficiency and reducing computational costs.

Collaborative Stance Detection via Small-Large Language Model Consistency Verification