The spread of fake news has emerged as a serious societal concern, driven largely by the explosive growth of multimedia content such as text, images, and videos on social media platforms. While numerous detection methods attempt to leverage both textual and visual modalities to identify misinformation, many rely on basic fusion strategies like feature concatenation or element-wise addition. These simplistic approaches often fall short in two critical aspects: they struggle to effectively capture the complex relationships between different data modalities (text and images), and they lack the ability to accurately detect subtle inconsistencies or contradictions between the narrative presented in the text and the accompanying visuals. As a result, their effectiveness in identifying deceptive content remains limited. To overcome the limitations of existing fake news detection methods, this research paper introduces a novel approach named as Trustify, a fuzzy logic-based hybrid framework designed to enhance both the accuracy and interpretability of fake news identification. The proposed model employs a Long Short-Term Memory (LSTM) network to capture contextual nuances in the text and introduce a GNN embedded Transform (encoding–decoding with object relationship mapping) to classify image content with high precision. The proposed system seamlessly integrates textual content, visual data (images), and their semantic alignment to form a more holistic and reliable detection strategy. The effectiveness of the framework is validated through comprehensive experiments on three benchmark datasets: Twitter, Buzzfeed, and PolitiFact, where it outperforms current state-of-the-art approaches across all key evaluation metrics, including accuracy, precision, recall, and F1-score.
Bharati et al. (Fri,) studied this question.