FNED: Integrated Learning-Based Fake News Inspection

This paper addresses the challenge of inadequate detection accuracy and generalization in false news detection amid the rapid spread of misinformation in the news environment, proposing FNED, an integrated learning-based false detection method. The method enhances detection performance by integrating multiple models; it employs a combination of three models—KNN, LSTM, and XGBoost—utilizing a stacking technique to leverage their predictions as features, followed by logistic regression as a meta-classifier for comprehensive evaluation. Augment the model's resilience across varying data distributions. Experiments utilize the public FakeNewsNet standard dataset, revealing that FNED enhances accuracy by approximately 1.3% relative to previous approaches, indicating superior detection capability and generalization performance.