The rapid proliferation of misinformation across social media platforms has emerged as a critical challenge, necessitating intelligent and interpretable detection mechanisms. Although transformer-based models such as BERT achieve high accuracy in textual analysis, they lack transparency and fail to capture temporal propagation dynamics. To address these limitations, this paper proposes a hybrid explainable framework that integrates Bidirectional Encoder Representations from Transformers (BERT) for semantic feature extraction with Spiking Neural Networks (SNNs) for modeling temporal engagement patterns. The proposed approach leverages the event-driven nature of SNNs to efficiently capture time-dependent user interaction signals, while SHAP-based explainability provides interpretable insights into both textual and temporal contributions. Experiments conducted on the FakeNewsNet dataset demonstrate that the proposed model outperforms baseline approaches, achieving an F1-score of 0.94 while maintaining improved computational efficiency. The results highlight the effectiveness of combining semantic understanding, temporal modeling, and explainability in a unified framework for reliable misinformation detection.
Karale et al. (Fri,) studied this question.