Political sentiment analysis using natural language processing on social media

Abstract

In this contemporary era , social media has become an essential component of daily life as a result of the extensive use of the internet. This paper explores sentiment analysis of political topics through social media comments. We collected a large dataset of over 14,000 political comments and applied advanced machine learning models such as logistic regression , linear support vector classification , random forest, decision tree classification , and naive bayes to evaluate expressed sentiments. Performance metrics , including accuracy , precision , recall , and F1 scores , were utilized to assess these models , with Linear SVC achieving the highest accuracy at 91.18% , closely followed by Logistic Regression at 90%. This research not only evaluates model performance on political sentiment data but also addresses data imbalance, presenting actionable insights into each algorithm’s suitability. Our study introduces a refined approach to political sentiment analysis by optimizing model selection for high accuracy and robustness, thus setting a foundation for effective political sentiment understanding on social media platforms.In this contemporary era , social media has become an essential component of daily life as a result of the extensive use of the internet. This paper explores sentiment analysis of political topics through social media comments. We collected a large dataset of over 14,000 political comments and applied advanced machine learning models such as logistic regression , linear support vector classification , random forest, decision tree classification , and naive bayes to evaluate expressed sentiments. Performance metrics , including accuracy , precision , recall , and F1 scores , were utilized to assess these models , with Linear SVC achieving the highest accuracy at 91.18% , closely followed by Logistic Regression at 90%. This research not only evaluates model performance on political sentiment data but also addresses data imbalance, presenting actionable insights into each algorithm’s suitability. Our study introduces a refined approach to political sentiment analysis by optimizing model selection for high accuracy and robustness, thus setting a foundation for effective political sentiment understanding on social media platforms.