<![CDATA[The mandatory policy of 10% ethanol blending in fuel oil (BBM E10), announced in October 2025, triggered a wide range of responses from the Indonesian public, particularly on the X social media platform. This study aims to analyze the sentiment of Platform X users towards the BBM E10 policy and evaluate the performance of a classification model built using the Multinomial Naïve Bayes algorithm. The method applied is Knowledge Discovery in Database (KDD), encompassing data selection, preprocessing, transformation, data mining, and evaluation stages. Data was collected through crawling using the Twitter API with keywords "ethanol" and "BBM E10" during September 2025 to January 2026, yielding 4,847 tweets. After removing duplicates, 2,385 records were labeled using a hybrid approach—combining lexicon-based automatic labeling with manual verification by a language expert. Feature transformation was performed using TF-IDF. Classification was conducted under three data splitting scenarios: 70:30, 80:20, and 90:10. Results showed that positive sentiment dominated with 1,034 tweets (43.35%), followed by negative with 989 tweets (41.47%), and neutral with 362 tweets (15.18%). The 90:10 scenario produced the best performance with an accuracy of 70.7%, precision of 69%, recall of 57%, and F1-Score of 54%. This study demonstrates that the Multinomial Naïve Bayes-based text mining approach is capable of automatically classifying public sentiment towards the BBM E10 policy with adequate accuracy.]]>
Copyrights © 2026
