Social Media Insights

Abstract-This research presents a comprehensive sentiment analysis system tailored for social media comments, aiming to classify user sentiments into positive, negative, or neutral categories. With Social media’s vast user engagement—over 1 billion unique users generating extensive comment data—there exists a significant opportunity to derive insights into public opinions. This study addresses challenges inherent in analyzing social media comments, including the high volume of data, diverse linguistic expressions, the use of slang, emojis, sarcasm, and the presence of spam. We leverage a constructed annotated corpus comprising 1500 citation sentences, which underwent rigorous data normalization to enhance quality and consistency. Six machine learning algorithms— Naïve Bayes, Support Vector Machine (SVM), Logistic Regression, Decision Tree, K-Nearest Neighbor (KNN), and Random Forest (RF)—were implemented for sentiment classification. The performance of these algorithms was evaluated using various metrics, including F-score and accuracy, demonstrating a correlation between sentiment trends and real-world events associated with specific keywords. This work contributes to the field of sentiment analysis by providing insights that can aid researchers in identifying quality research papers and understanding user attitudes towards video content.