Enhancing fake news detection for Sustainable Supply Chain Management using DistilBERT-based multi-stacked LSTM approach

Abstract

In Sustainable Supply Chain Management (SSCM), credible information is vital to gain stakeholder trust for sustainable development. However, misinformation and fake news can lead to poor decisions, reputational damage, and reduced performance. This study proposes a DistilBERT-based multi-stacked Long Short-Term Memory (LSTM) model to improve fake news classification using contextual embeddings in SSCM. Feature extraction is performed using Word2Vec and TF-IDF within a High Feature Extraction (HFE) framework. The best subset is selected using the Least Absolute Shrinkage and Selection Operator (LASSO). Evaluated on the WELFake dataset, the proposed model achieves 99.82% accuracy, outperforming existing methods in detecting SSCM-related fake news.