A Scalable Machine Learning Model for Enhancement of Fake News Detection
The rise of fake news has become a critical issue in the digital era, especially on social media platforms where information spreads rapidly and widely. Fake news, defined as deliberately misleading or false information presented as factual news, poses threats to public health, democracy, and economic stability. While machine learning offers scalable solutions for detecting such content, many existing models face challenges related to scalability, portability, and data quality. This study addresses these limitations by developing a scalable machine learning model to enhance fake news detection across digital platforms. The ISOT Fake News Dataset was employed, and extensive data preprocessing was conducted, including text normalization, lemmatization, and TF-IDF vectorization. Four machine learning algorithms: Logistic Regression, Naive Bayes, Support Vector Machine (SVM), and Random Forest were implemented and evaluated. Exploratory Data Analysis (EDA) was also introduced to uncover data patterns and guide model development. The results showed that SVM and Random Forest achieved superior accuracy (1.00), outperforming Logistic Regression (0.99) and Naive Bayes (0.93). The findings suggest that ensemble-based and kernel-based models are more robust for high-dimensional text classification tasks. Moreover, EDA contributed to improved model interpretability and performance. In conclusion, this study demonstrates that a carefully engineered, scalable approach can significantly improve fake news detection accuracy. Future work will explore deep learning models, cross-lingual datasets, and real-time deployment frameworks to enhance model generalizability and responsiveness.
Download Paper