Multilingual Fake News Detection Using Hybrid Model for Enhanced Computational Efficiency and Performance in Low-Resource Languages

Pippalla, Srinadh (2025) Multilingual Fake News Detection Using Hybrid Model for Enhanced Computational Efficiency and Performance in Low-Resource Languages. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

The issue of fake news in the digital era, especially on a global and multilingual landscape, poses a serious threat to information integrity, especially in low-resource linguistic contexts. The conventional fake news detection tools are usually developed based on high-resource languages and low resource languages are underrepresented because of the lack of annotated data and computational resources. This study seeks to solve the urgent problem of an effective and scalable multilingual fake news detection system that serves well in both high- and low-resource languages. The research suggests hybrid deep learning model which integrates a state-of-the-art transformer model, DeBERTa, which has demonstrated an ability to represent multilingual data with a Bidirectional LSTM network to improve the temporal comprehension and classification accuracy. A multilingual corpus of seven languages was compiled and cleaned. To set benchmarks, the use of classical machine learning models (Logistic Regression, Naive Bayes, Random Forest) and deep learning baselines (GRU, CNN) was implemented. The experimental findings proved that the suggested DeBERTa + LSTM hybrid model was much better than baselines, showing a high overall accuracy of 97%, and performing especially well when dealing with low-resource languages. The hybrid mode did not only improve the accuracy of classification but also maximized the computational performance by freezing the transformer backbone, which can be used in practice in resource-limited settings. The study is an addition to the emerging discipline of multilingual NLP as it introduces a scalable, powerful solution to the problem of misinformation detection. It opens up the path towards future work in cross-lingual fake news classification, low-resource language adaptation, and deployment-ready hybrid architectures.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Niculescu, Hamilton UNSPECIFIED
Subjects:	P Language and Literature > P Philology. Linguistics > Computational linguistics. Natural language processing Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning T Technology > TK Electrical engineering. Electronics. Nuclear engineering > Multimedia Communications Z Bibliography. Library Science. Information Resources > ZA Information resources > ZA4150 Computer Network Resources > The Internet > World Wide Web > Websites > Online social networks T Technology > TK Electrical engineering. Electronics. Nuclear engineering > Telecommunications > The Internet > World Wide Web > Websites > Online social networks
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	02 Jul 2026 14:59
Last Modified:	02 Jul 2026 14:59
URI:	https://norma.ncirl.ie/id/eprint/9449

Actions (login required)

View Item