Request Correction

Credibility Assessment for Arabic News on the Gaza War: A Hybrid Neural-Symbolic Pipeline

While misinformation has long circulated online, the Gaza conflict has intensified its visibility and spread across news websites, online portals, and social media, complicating the credibility and long-term curation of conflict-related Arabic records, including historical accounts and written testimonies. This work proposes a hybrid framework for Arabic fake news detection that combines interpretable linguistic cues with contextual semantic representations. The approach integrates fuzzy logic-based handcrafted features capturing exaggerated and sensational linguistic patterns, AraBERT contextual embeddings for semantic understanding, and a CNN-based text feature extractor for local textual patterns. These complementary features are combined into a unified representation for downstream classification. Multiple machine learning and deep learning classifiers are evaluated to identify the most effective detection model. The resulting system is deployed as a real-time web browser plugin, enabling users to automatically assess the credibility of Arabic news content during browsing