Author(s): Nikita Garg and Pritam Singh Negi

Abstract: The widespread dissemination of fake news online poses a serious threat to reliable information access, highlighting the need for efficient detection methods. This study introduces a novel hybrid feature selection framework that integrates the Grey Wolf Optimizer (GWO) and Artificial Bee Colony (ABC) algorithms to improve fake news classification. The framework combines GWO’s global search capabilities with ABC’s local exploitation to reduce feature redundancy while retaining the most informative textual attributes. Using Term Frequency-Inverse Document Frequency (TF-IDF), an initial set of 5,000 features is extracted from a publicly available news dataset and reduced to 2,496 optimized features through the hybrid approach. These features are evaluated using five machine learning classifiers, with XGBoost achieving the highest accuracy of 90% along with balanced precision, recall, and F1-score, demonstrating the effectiveness and stability of the method. Compared to conventional or deep learning models, this approach offers a lightweight and computationally efficient solution without sacrificing performance. The proposed framework provides a new direction in metaheuristic-based feature optimization for textual fake news detection and can be extended to multimodal, multilingual, and real-time applications, offering a practical tool for mitigating misinformation.

DOI: 10.33545/27076571.2025.v6.i2d.208

Pages: 287-292 | Views: 65 | Downloads: 31

Download Full Article: Click Here