détection des rumeurs dans les réseaux publics

Abstract

The rapid development of social media has facilitated the exchange of large amounts of data but has also accelerated the spread of false information. Several studies have focused on rumor detection by primarily analyzing the textual content of messages. However, visual content, particularly images, remains largely underutilized. Yet, images are ubiquitous on social media, and their use is essential for a comprehensive analysis of rumors. In this study, we present a synthesis of current research on rumor classification, summarizing the key steps of this process and the approaches used to study it. The objective of our work is to develop an automatic rumor detection system using deep learning based on the recurrent neural network (RNN) model to recognize spam in the KAGGLE database. We evaluated the capabilities and performance of our system by testing it on a test dataset after integrating and validating the RNN model on the preprocessed data. The results show an accuracy of approximately 99.89% with a negligible error rate.