This paper proposes a novel hybrid fake news detection system combining a BERT-based model with a LightGBM model. The system's performance is compared against four other classification approaches using various word embedding techniques across three real-world datasets. Evaluations using headline-only and full-text news content demonstrate the superior performance of the proposed method compared to state-of-the-art methods.
Publisher
Springer
Published On
May 01, 2023
Authors
Ehab Essa, Karima Omar, Ali Alqahtani
Tags
fake news detection
BERT model
LightGBM
classification approaches
word embedding
evaluation
real-world datasets
Related Publications
Explore these studies to deepen your understanding of the subject.
