Integrating Explainable AI with Enhanced Ensemble Models for Accurate and Transparent Fake News Detection in OSN’s

Nowadays, identifying and mitigating fake news creates challenges, particularly due to the widespread dissemination of text-based content over internet sites, online social networking platforms, and emails. While conventional ways of detecting fake news can be quite successful, they frequently fall behind the latest tactics. The developments in machine learning and deep learning algorithms have greatly enhanced the performance of fake news-detecting frameworks. This work develops a reliable and understandable method to identify fake news efficiently by combining XAI with ensemble learning and employing different techniques to increase efficacy. Nevertheless, user dependability is hampered by these products’ opaque and black-box design. Explainable AI has surfaced in subsequent actions to improve people’s comprehension of artificial intelligence conclusions. We utilized four classifiers to perform testing and training: decision tree, support vector machine, and logistic regression, respectively. Three different fake news datasets were balanced and combined to lessen overfitting. Compared to individual classifiers, the Random Forest (RF)-based stacking ensemble approach yielded the greatest results, with 98% recall, 96% accuracy, and 96% F1-score. Using XAI’s interpretability, the approach explains its categorization facts, revealing unseen techniques for recognizing fake news.