<![CDATA[The phenomenon of fatherlessness in Indonesia is on the rise due to divorce, patriarchal culture, and the lack of fatherly involvement. Social media, particularly the X platform, has become the primary space for sharing opinions and experiences related to this issue. This study analyzes public sentiment related to fatherlessness and compares the performance of two sentiment classification methods: Support Vector Machine (SVM) and IndoBERTweet, using two testing scenarios. Data was collected through crawling from the X platform and manually labeled into four sentiment categories: positive, negative, neutral, and others. Evaluation was conducted using a confusion matrix, classification report, and error analysis. The results show that IndoBERTweet achieved the highest accuracy of 0.92, while SVM reached 0.86. Error analysis plays a crucial role in identifying misclassification patterns, particularly in texts with sarcasm and ambiguity, which make it challenging for models to distinguish sentiments with similar contexts or tones across labels. ]]>
Copyrights © 2026
