A multi-modal approach for recognizing fake news and influential nodes in spreading them using deep learning and network analysis

Rumors may spread swiftly and cause a lot of harm on social media. Finding the main nodes that fuel rumor's spread is necessary to lessen their impact. This study proposes a new technique that combines deep learning with graph-based clustering techniques to effectively detect rumor sources and predict the paths of spread in social networks. "Detection of rumors" and "Identifying the influential nodes in the spread of rumors" are the two primary steps of the algorithm that is being described. A criterion based on the rate of rumors spreading is used in the first phase of the suggested method to determine the significance of each connection in the social network graph. The characteristics of the contents are then described using the GloVe technique. After this stage, rumors are separated from other content using a deep neural network. In the second phase of the suggested approach, communities inside the social network structure are found using a novel clustering methodology, and each user's significance in spreading rumors is determined. The accuracy, precision, and processing time criteria have been used to assess the suggested method's effectiveness in determining the routes taken by rumors as they spread. In the trials, real social network databases were employed. The obtained findings demonstrate that the suggested method may effectively identify the path taken by rumors as they travel through social networks. The results of the case study also showed that the proposed method could recognize target data by average accuracy of 99% and achieve the precision of 0.99 in recognizing influential nodes.