<![CDATA[The development of social media has led to a significant increase in the number of consumer reviews of various types of products, including coffee products. To help manufacturers understand consumer satisfaction levels more efficiently, sentiment analysis is a relevant method because it is able to identify opinions automatically. This study compares the performance of two widely used algorithms, namely Support Vector Machine (SVM) and Multinomial Naive Bayes (MNB), in predicting sentiment on consumer reviews related to coffee products on social media. The dataset was analyzed through the stages of text cleanup, TF-IDF transformation, and label encoding process. Both models are developed using a uniform pipeline with consistent parameters to ensure an objective performance comparison. The results show that SVM algorithms with linear kernels produce the highest accuracy compared to Naive Bayes. In addition, a confusion matrix is applied to evaluate the accuracy of predictions in each sentiment category. These findings confirm that SVM is more effective in short-text-based sentiment analysis tasks, such as product reviews on social media platforms.]]>
Copyrights © 2026
