Towards Development of Machine Learning Models for Fake News Detection and Sentiment Analysis
摘要
The surging popularity of social media has yielded an abundance of textual data, greatly enhancing its potential for analysis. Users freely express their opinions and sentiments around the clock, irrespective of time or place. Moreover, the widespread dissemination of misinformation on social media platforms has reached alarming levels. This study focused on the development of machine and deep learning-based models, namely Random Forest (RF), Naïve Bayes (NB), Logistic Regression (LR), a Majority Voting-based Classifier, and Bidirectional Long Short-Term Memory (Bi-LSTM). These five models are designed to analyze text data for the identification of fake news and the classification of customer review sentiments. Utilizing an 70–30% train-test split, the Bi-LSTM model demonstrated an impressive accuracy of 97.11%, 92.01% in the classification of fake news and customer review sentiments, respectively.