Thread User Sentiment Analysis Based on Text Using LGBM, SVM, and Logistic Regression Algorithm

Social media is growing rapidly as a medium for online communication. Through social media, users can share opinions, stories or other things. The messages shared tend to have various meanings, such as positive, negative or neutral. The text shared sometimes has a negative meaning that can disturb other users. So based on this, this study will conduct a Text-Based Thread User Sentiment Analysis using Light Gradient Boosting Machine (LGBM), Support Vector Machine (SVM), and Logistic Regression. Our comparative investigation using a large dataset of 37,000 annotated text samples from the Thread application shows that SVM outperforms other methods with an accuracy of 89.24%. The accuracy rates for LGBM and logistic regression are 83.89% and 86.91%, respectively. One of the evaluation metrics is precision, which consistently shows superior SVM performance. These findings indicate that Support Vector Machine (SVM) is the most efficient algorithm for sentiment analysis in textual data, indicating its reliability and suitability.