FAKE NEWS DETECTION MODEL USING ENSEMBLE BI-DIRECTIONAL LSTM AND SUPPORT VECTOR MACHINES

Abstract

In a time where information is quickly exchanged across numerous digital platforms, the emergence of fake news has become a serious concern. Creating reliable and effective false news detection methods is necessary to meet this problem. By combining the benefits of Support Vector Machine (SVM) classifiers and Bi-Directional Long Short-Term Memory (Bi-LSTM) networks, this study proposes a novel method for improving fake news identification. The suggested model makes use of Bi-LSTM networks' capacity to recognize sequential dependencies in textual material. The model successfully captures context and semantic distinctions that are crucial for distinguishing between real news and fake news by processing incoming text in a bidirectional manner. Support Vector Machines are also used to take advantage of their effectiveness in dividing feature spaces and defining difficult decision boundaries. The experimental data show that the proposed ensemble technique performs better in terms of F1-score, recall, accuracy, and precision. A potential first step in curbing the spread of misinformation is the model's ability to differentiate between authentic content and fake news. The ensemble framework's adaptability and performance make it a powerful weapon in the ongoing fight against the spread of false information in the digital sphere as fake news production techniques develop.