A Multi-Label Fake News Detection Framework Using Undersampling and Transformer-Based Deep Learning Models

View at publisher Summarise with VIRAT
Authors: Pratyush Ranjan Mohapatra, Jayadeep Pakala, Nagagopiraju Vullam, Vunnava Dinesh Babu, Naladi Ram Babu, P. Chandra Sekhar Reddy
Cited by: 0
References: 12

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The rapid spread of misleading and fabricated information across social media platforms has intensified the need for robust fake news detection systems. Traditional binary classification methods are insufficient for real-world scenarios where news items often receive multiple credibility judgments. To address this challenge, this paper proposes a multi-label fake news detection framework that integrates undersampling strategies with advanced transformer-based deep learning models. The dataset used in this study consists of tweets annotated with multiple credibility ratings, enabling multi-label classification through annotator-based label vectors. Two undersampling techniques—One-vs-Rest (OVR) and MultiLabel Random Undersampling (MLRU)—are employed to mitigate the effects of label imbalance and improve detection of minority credibility classes. Classical machine-learning algorithms, deep learning architectures, and transformer models including BERT and RoBERTa are evaluated on both undersampled datasets. Experimental results demonstrate that transformer models consistently outperform traditional approaches, with RoBERTa achieving the highest performance across all evaluation metrics. Furthermore, the application of MLRU significantly improves class-wise recall in minority categories. These findings highlight the effectiveness of combining undersampling with transformer architectures for multi-label fake news identification in highly imbalanced realworld settings.

Fields of Study & Subjects

Misinformation and Its Impacts Spam and Phishing Detection Hate Speech and Cyberbullying Detection

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
1074–1079
Publication date
21 January 2026
Language
EN