Information Veracity Detection on Social Media using Neural Networks
Abstract
Online social media platforms are a hub of popular rumors, therefore demystifying them has paramount significance. To identify rumors, general strategies depend on hand-crafted apps to employ machine learning algorithms that demand overwhelming manual work. Challenged with a doubtful argument, people challenge its truthfulness by posting different indications over time that lead to long-distance reliance. This study showcases metric based comparisons of recurrent neural network approaches that classify representations of online blogging/tweeting activities to determine rumors. The approach hinges on comparing recurrent neural networks (RNN) for the detection of latent patterns that acquire fluctuations in the qualitative knowledge of the related posts over time. Furthermore, the advanced recurrent units along with extra secret layers help improve overall performance. Results show that RNN-based approach discovers rumors more rapidly and precisely than more standardized methods.