Tweet Sentiment Analysis with Support Vector Machine (SVM) Algorithm for PT. XYZ Digital Strategy

Abstract

The development of the digital era has encouraged the use of social media as a strategic tool for promotion and information dissemination, including in the publishing industry. PT. XYZ utilizes Twitter to raise awareness of literacy issues. This study analyzes public sentiment towards literacy through tweets with the hashtags #literasi, #literasiindonesia, and #penerbitbuku using the Support Vector Machine (SVM) algorithm. The research stages include scraping, preprocessing, labeling, TF-IDF vectorization, and evaluation with a confusion matrix. From the 466 tweets analyzed, a balanced distribution of positive and negative sentiments was obtained. The model produced an accuracy of 47.87%, precision of 47%, recall of 57%, and an F1-score of 51%. These results are lower than those of Afrillia et al. (2022), who achieved an accuracy of 70.8%, and Widyanto et al. (2023), who obtained 80.41% with an RBF kernel. These differences confirm the limitations of SVM on small datasets and informal language. This study contributes by showing the potential and limitations of SVM in analyzing literacy on social media. These results also emphasize the need for further research with larger datasets and advanced methods such as ensemble learning and deep learning (LSTM, BERT).