Leveraging Large Language Models for Misinformation Detection: A Focus on Public Health Misinformation on Social Media

Document Type

Conference Proceeding

Source of Publication

Lecture Notes in Networks and Systems

Publication Date

10-1-2025

Abstract

In recent years, especially following the COVID-19 outbreak, the spread of misleading public health information on social networking sites has gained new significance and scale. Misinformation campaigns now pose a serious threat to the credibility of information, eroding public trust in authorities and healthcare institutions, and underscoring the urgent need for effective detection. This paper introduces a system specifically designed to detect public health misinformation campaigns. Leveraging Large Language Models such as Llama 3.1 8B and Natural Language Processing (NLP), our system analyzes linguistic patterns to accurately interpret tweet context and distinguish misinformation from factual content. This approach empowers individuals, organizations, and policymakers to make well-informed decisions, fostering social cohesion and strengthening public trust. Our system is powered by a large, curated dataset compiled from 17 representative fake news datasets collected from three trusted sources. These datasets include both real and fake news, covering general topics as well as specific public health issues, particularly COVID-19, and are gathered from social media platforms like Twitter and various news outlets. Leveraging the Llama 3.1 8B Instruct LLM, our model achieves a remarkable 99% accuracy in detecting public health misinformation.

ISBN

[9789819692477]

ISSN

2367-3370

Publisher

Springer Nature Singapore

Volume

1539 LNNS

First Page

561

Last Page

573

Disciplines

Communication | Computer Sciences

Keywords

Fake news, LLM, Misinformation, NLP, Social media, Twitter

Scopus ID

105020188345

Indexed in Scopus

yes

Open Access

no

Link to Full Text

Share

COinS