A Review on Machine Learning Techniques for Fake News Detection

Abstract—Spreading false news online has become a gravethreat to public stability, political life, and trust. Conventionalcontent moderation measures prove inadequate to contain thevelocity and volume with which disinformation spreads. MachineLearning (ML) has been an invaluable aid in the automation offake news identification by learning patterns and cues from largeamounts of text data. This review paper provides an overviewof most of the ML methods employed for the detection offake news, from the simple algorithms Naive Bayes, SupportVector Machines, and Decision Trees to the recent deep learningmethods such as LSTM, CNN, and transformer-based modelssuch as BERT. This work also encompasses feature extractionmethods like TF-IDF, word embeddings, and stylometric features,benchmarking datasets, and evaluation metrics used in theliterature. The work provides the advantages and limitationsof existing methods and highlights open problems such asdataset generalizability, model explainability, and adversarialrobustness Lastly, the review outlines future directions, such asthe development of hybrid models, real-time detection systems,and ethics. The intention of this paper is to be a starting groundfor researchers as well as practitioners in creating improved andmore efficient fake news recognition systems.Index Terms—Fake News Detection, Machine Learning, NaturalLanguage Processing, Text Classification, Deep Learning,Transformer Models, Social Media Misinformation, NLP, NewsVerification