ExFake: Towards an Explainable Fake News Detection Based on Content and Social Context Information

Fake news detection requires multiple information sources: ExFake combines content analysis, user credibility scores, and fact-checking website data rather than relying on any single approach, demonstrating that a comprehensive strategy is more effective than examining posts in isolation.

User credibility matters as much as content: The system assigns credibility scores to social media users based on their historical behavior, recognizing that "OSN users are one of the main culprits for spreading fake news" and that who shares information is as important as what they share.

Explainability helps users make better decisions: Unlike black-box detection systems, ExFake includes an "Explainable AI assistant" designed to help social media users develop critical thinking skills when encountering questionable information, shifting from passive detection to active user education.

Trusted external sources strengthen detection: The system incorporates data from fact-checking websites and named entity information, suggesting that integrating established authoritative sources into detection systems improves accuracy and reliability.

Explainable systems outperform traditional methods: The experimental results show that ExFake "significantly outperforms other baseline methods," indicating that transparency in how fake news is identified doesn't compromise—and may actually improve—detection effectiveness.