1. Arafakedetect: enhancing fake health news detection with ensemble learning on AraCovidVac.
- Author
-
Mahmoud, Samar, Aboutabl, Amal Elsayed, and Mohamed, Ensaf Hussein
- Abstract
The broad availability and use of the internet have dramatically amplified the already existing problem of false information, particularly when it comes to health-related news. This phenomenon became a major concern during the COVID-19 pandemic, where false and fraudulent information spread rapidly, posing a significant threat to public health. The COVID-19 pandemic served as a powerful illustration of this phenomenon, as false and fraudulent health information spread rapidly across online platforms, creating confusion and jeopardizing public health efforts. This highlights the urgent need for solutions to address the spread of fake information, particularly in the area of health, where the consequences of false information can have a direct and potentially devastating impact on individuals and communities. To solve this issue, we propose an ensemble and stacking model that leverages Support Vector Machines (SVM), Arabic Bidirectional Encoder Representations from Transformers (AraBERT), Bidirectional Long Short-Term Memory (Bi-LSTM), and Logistic Regression (LR). Our model leverages the strengths of different machine learning techniques to mitigate the spread of harmful false information. To tackle this issue, we use the largest manually annotated Arabic dataset, ArCovidVac, focusing on the COVID-19 vaccination discourse, and we leverage an additional dataset for COVID-19 to generate additional training data for our fake news detection model. The experimental results demonstrate that our framework performs better than previous state-of-the-art approaches. Our proposed model achieved an accuracy, exceeding the performance of AraBERT. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF