Twitter Sentiment Analysis on Women Violence using Machine Learning Techniques

This research paper delves into the prevalent practice of sentiment analysis on social media platforms, particularly Twitter, where users actively express their opinions on diverse topics. Focusing specifically on women’s issues, with an emphasis on violence against women, the study extracts tweets from the hashtag #yesallwomen during the period from January to May 2020. Employing four distinct algorithms—Naïve Bayes, Random Forest, Logistic Regression, and Support Vector Machine—the research rigorously evaluates their performance using precision, recall, and F1 Score metrics. The study will show that the Support Vector Machine classification exhibits exceptional efficiency, achieving a notable accuracy of 97.98%. Analysis of positive tweets (46.49%) suggests a commendable awareness among women about their rights, while the prevalence of neutral tweets (40.75%) indicates potential areas for improvement. The research not only highlights the practicality of employing topic- modeling methods for scrutinizing violence against women data on Twitter but also proposes future extensions of the work. These include expanding the sentiment analysis to incorporate additional hashtags such as Metoo and TimesUp, leveraging a broader range of classification algorithms for comparative analysis, and thereby refining the understanding of public sentiment on women’s issues.