Comprehensive Sentiment Analysis of Tweets Us...

Comprehensive Sentiment Analysis of Tweets Using Supervised Machine Learning Approach

Author Name : Jagram, Dr. Shobhit Srivastava, Piyush Rai, Nidhi Prasad

DOI: https://doi.org/10.56025/IJARESM.2024.1209240592

ABSTRACT With the rise of technology, micro blogging sites like Twitter are increasingly used to share opinions on current topics, products, services, and events. This study developed a feature-based Twitter Sentiment Analysis (TSA) framework using supervised machine learning, incorporating sophisticated negation handling and a knowledge based Tweet Normalization System (TNS). We created three real-time Twitter datasets (#Demonetization, #Lockdown, #9pm9minutes) and used a benchmark dataset (SemEval-2013) to evaluate the framework's effectiveness. Our analysis employed features such as lexicon-based, part-of-speech, n-grams, and negation features, and tested classifiers including Support Vector Machine (SVM), Decision Tree (DTC), and Naive Bayes (NB). SVM generally performed best, except for the #9pm9minutes dataset where DTC excelled. Our SVM model also outperformed the top entry in the SemEval-2013 competition.