A COMPARATIVE STUDY OF MACHINE LEARNING TECHNIQUES FOR REAL-TIME SENTIMENT ANALYSIS USING WEB SCRAPING

Real-time sentiment analysis is an important task for understanding public opinions, emotions, and attitudes expressed through online content. This study presents a machine learning-based Real-Time Sentiment Analysis System with Web Scraping to classify textual data into positive and negative sentiments. The proposed system uses the Sentiment140 dataset, which contains Twitter messages labelled according to their sentiment polarity, for training and evaluating classification models. Three machine learning algorithms, namely Naïve Bayes, Linear Support Vector Machine (SVM), and Stochastic Gradient Descent (SGD) Classifier, are considered for comparative evaluation. The methodology involves text preprocessing, removal of unnecessary characters and URLs, transformation of textual data into numerical features using Term Frequency–Inverse Document Frequency (TF-IDF), model training, sentiment prediction, and performance evaluation. Web scraping is incorporated to collect relevant textual content from online sources for sentiment analysis, where applicable. The performance of the models is evaluated using metrics such as accuracy, precision, recall, F1-score, and confusion matrix to understand their classification capabilities. The experimental results obtained for the Linear SVM and SGD models show accuracies of 80.49% and 77.60%, respectively, under their respective experimental settings. The proposed system provides a framework for analyzing online text and identifying positive and negative opinions efficiently. This approach can support applications such as product review analysis, social media opinion monitoring, and customer feedback analysis. The comparative study highlights the applicability of machine learning techniques in sentiment classification and provides a basis for selecting suitable models for real-time sentiment analysis applications.