Handling Imbalanced Classes for Model Training in Fake News Detection

Abstract

With the widespread dissemination of news on social media platforms, the propagation of fake news has become a pressing concern. Detecting fake news is crucial to maintaining the integrity of information shared across social networks. This paper presents a comprehensive investigation into the detection of fake news on social media, focusing on the collection of data from both reliable and unreliable sources To build an effective fake news detection system, a diverse dataset encompassing both reliable and unreliable sources is collected.  This data collection strategy ensures a comprehensive representation of the information landscape present on social media platforms.The implementation of the bidirectional LSTM with an attention layer is a powerful approach that has shown promising results in various natural language processing tasks, including text classification and sentiment analysis. Its effectiveness lies in its ability to leverage both directional information and attention-driven focus, allowing the model to better understand and interpret the nuances of the input sequence. Calculate the class weights using the inverse of the importance factors to give proper weight to each label. Balancing has been tried with the class weight method the system has given almost 60% accuracy.