Analisis Sentimen Masyarakat Terhadap Kinerja Presiden Indonesia Joko Widodo Periode Kedua Menggunakan Metode Naïve Bayes dan SVM

Abstract

The advancement of information technology, particularly social media, has transformed the way the public expresses opinions on public issues, including the performance of the president. This study aims to analyze public sentiment regarding the performance of the Indonesian President during his second term using two text classification methods: Naïve Bayes and Support Vector Machine (SVM). The dataset consists of 1,003 tweets collected from social media platform X between September 2023 and September 2024. Prior to classification, the data underwent preprocessing steps such as cleaning, normalization, case folding, stopword removal, and stemming. The classification results revealed that 57.83% of tweets expressed negative sentiment, 34.40% positive, and 7.78% neutral. Negative sentiments were predominantly associated with issues such as price hikes, controversial policies, and allegations of corruption, whereas positive sentiments related mainly to infrastructure development and social assistance programs. Performance evaluation indicated that the SVM method achieved a higher accuracy of 71.6%, outperforming Naïve Bayes, which achieved 65.2% accuracy. This study concludes that social media serves as an effective data source for capturing broad public opinion, and that SVM is a more effective classifier than Naïve Bayes for sentiment analysis of social media text data.