Fake News Detection Using Natural Language Processing and Machine Learning Techniques

In an era dominated by digital information and rapid technological advancements, the spread of false information, commonly known as fake news, has emerged as a significant problem. This study offers a comprehensive technique for identifying false news, including dataset collection, text preprocessing, feature extraction, and Machine Learning algorithms. The dataset used includes both true and fake news stories that were gathered from reliable sources. Text preprocessing techniques such as Stop word removal, parts of speech tagging, Tokenization, Stemming, and Lemmatization are used to clean and prepare the textual data. In this study Feature extraction techniques like TF-IDF Vectorization and Bag of Words are used to transform the text into a numerical representation. The paper evaluates the performance of ML classifiers, including Logistic Regression, Decision Tree, Support Vector Machine, Naïve Bayes, Gradient Boosting, and Random Forest, in detecting fake news. Notably, applying TF-IDF Vectorization to Decision Tree and Gradient Boosting classifiers resulted in high accuracy ratings. Similarly, when combined with Bag of Words, Logistic Regression, Decision Tree, and Gradient Boosting all achieved outstanding accuracy scores.