Empirical Analysis on Fake News Detection Using Feature Extraction and Feature Optimization Techniques

With the rapid growth of technology and social media, fake news has become a major issue today. Detecting and preventing fake news is becoming increasingly difficult and is a significant problem that requires early detection. The spread of fake news on social media is a growing concern, as it poses a threat to public opinion and trust in news media. Researchers have developed various methods and algorithms to identify and classify fake news, including supervised and unsupervised machine learning approaches and ensemble methods. This study aims to compare and analyze fake news detection methods based on feature optimization algorithms such as swarm, firefly, and particle swarm optimization. The study compares feature extraction and optimization techniques with deep learning classifiers to validate the efficiency of metaheuristic algorithms. Features vectors are extracted using advanced techniques such as TF-IDF and sentence transformers like BERT and RoBerta, and fitness vectors are extracted via metaheuristic swarm intelligence (SI) algorithms. Convolutional neural networks (CNNs) are utilized as classifiers to improve classification accuracy. The LIAR dataset is used to train and test the classifier, and the results show that incorporating swarm intelligence algorithms and advanced feature extraction techniques can significantly improve the classification of fake news.