Comprehensive Analysis of Machine Learning Models and Hybrid Approaches for Fake News Detection

The false circulation of fake news across digital platforms has highlighted the urgent need for reliable detection systems to preserve the accuracy of information. The usefulness of different ML methods for detecting fake news is investigated and compared in this paper. We evaluated the performance of eight widely used classifiers: KNN, Naive Bayes, Random Forest Classifier, Logistic Regression, SVM, GBC, Neural Networks, and AdaBoost. The models are trained and tested on a balanced dataset using standard text preprocessing techniques. To determine which algorithm provides the most accurate classification of fake and real news, their precision, recall, F1-scores, and accuracy are evaluated. Our findings highlight the strengths and limitations of each model, offering insights into the most effective techniques for mitigating the spread of fake news. The results suggest that ensemble-based methods and deep learning approaches show significant potential in achieving higher accuracy rates, while traditional classifiers may fall short in handling complex patterns in the data.