This research introduces EffiARA, a novel framework for efficiently assessing annotator reliability to improve misinformation detection on social media.

Key insights:

• Applies natural language inference techniques to knowledge-based misinformation detection
• Develops methods to identify reliable annotators and weight samples accordingly
• Creates a more robust detection system by accounting for annotator quality
• Addresses the critical security challenge of rapidly spreading misinformation

Why it matters: This approach could significantly enhance automated security systems for social media platforms by improving the accuracy of misinformation detection while reducing the manual effort required in current processes like community notes.

Efficient Annotator Reliability Assessment and Sample Weighting for Knowledge-Based Misinformation Detection on Social Media