Semantic knowledge graph fusion for fake news detection: Unifying content-based features and evidence-based analysis in the COVID-19 infodemic.
In the era of digital communication, the rapid spread of information has brought both benefits and challenges. While it has democratized access to knowledge, it has also led to an increase in fake news, with significant societal repercussions. The COVID-19 pandemic has exacerbated this issue, result...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Public Library of Science (PLoS)
2025-01-01
|
| Series: | PLoS ONE |
| Online Access: | https://doi.org/10.1371/journal.pone.0321919 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In the era of digital communication, the rapid spread of information has brought both benefits and challenges. While it has democratized access to knowledge, it has also led to an increase in fake news, with significant societal repercussions. The COVID-19 pandemic has exacerbated this issue, resulting in what the World Health Organization has termed an "infodemic." In light of this, developing effective methods for detecting fake news is of paramount importance. In this paper, we introduce a novel approach that integrates knowledge graphs and Named Entity Recognition (NER) based on a biomedical language model to address the challenge of fake news detection. Our method aims to enhance detection accuracy by combining content analysis with entity-level insights. Our approach involves three key components. First, content analysis uses a contextual language model to capture the semantic context of the content, enabling the extraction of meaningful insights essential for identifying fake news. Second, the NER component, built on a biomedical language model, precisely identifies and categorizes entities within the content, offering a deeper understanding crucial for detecting misinformation in the biomedical domain. Finally, entity integration employs knowledge graph embeddings to transform identified entities into a format that facilitates enhanced processing and detection. By blending these components, our method creates a unified representation of the content, incorporating both semantic context and entity-based insights. This comprehensive approach significantly improves the accuracy of fake news detection. Our extensive experiments demonstrate the effectiveness of this method, particularly in the early identification of false information. The results underscore the potential of our approach as a powerful tool in combating misinformation, particularly in critical areas such as public health. |
|---|---|
| ISSN: | 1932-6203 |