ANALISIS SENTIMENT MASYARAKAT INDONESIA PADA MEDIA SOSIAL TERHADAP ISU IJAZAH PALSU MANTAN PRESIDEN MENGGUNAKAN ALGORITMA BERBASIS TRANSFORMER (BERT)
DOI:
https://doi.org/10.54314/jssr.v8i3.4209Abstract
Abstract: Social media has become one of the main channels for people to express their opinions on various social and political issues. This study aims to apply the BERT model in analyzing the issue to be analyzed to determine the public's reaction to the tendency of public sentiment, whether it is positive, negative, or neutral. BERT is very superior in understanding the meaning of sentences in depth, including in sentiment analysis tasks. The BERT evaluation shows that Negative sentiment with precision reaches 79%, recall 84%, and f1-score reaches 81%, for Neutral sentiment precision reaches 50%, recall reaches 45% and f1-score reaches 47%, positive sentiment precision reaches 66%, recall 61%, and f1-score reaches 63% and accuracy reaches 72%. Based on the overall results, the model produces an accuracy rate of 72%, which indicates that most of the sentiment predictions match the actual labels.
Â
Keywords: BERT, Sentiment.
Â
Abstrak: Media sosial telah menjadi salah satu saluran utama bagi masyarakat untuk menyampaikan opini terhadap berbagai isu sosial dan politik. Penelitian ini bertujuan untuk menerapkan model BERT dalam menganalisis isu tersebut untuk dianalisis guna mengetahui reaksi masyarakat pada kecenderungan sentimen publik, apakah bersifat positif, negatif, atau netral. BERT sangat unggul dalam memahami makna kalimat secara mendalam, termasuk dalam tugas analisis sentimen. Evaluasi BERT menunjukan bahwa sentimen Negatif dengan preccision mencapai 79%, recall 84%, dan f1-score mencapai 81%, untuk sentimen Netral precission mencapai 50%, recall mencapai 45% dan f1-score mencapai 47% , sentimen positif preccision mencapai 66%, recall 61%, dan f1-score mencapai 63% serta accuracy mencapai 72%. Berdasarkan hasil secara keseluruhan, model menghasilkan tingkat akurasi sebesar 72%, yang menunjukkan bahwa sebagian besar prediksi sentimen sesuai dengan label sebenarnya.
Â
Kata kunci: BERT, Sentimen.
Downloads
References
Devlin, J., Chang, M. W., Lee, K., & Toutanova, K. (2019). BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. arXiv preprint arXiv:1810.04805.
Kouloumpis, E., Wilson, T., & Moore, J. (2011). Twitter sentiment analysis: The good the bad and the OMG!. Proceedings of the Fifth International Conference on Weblogs and Social Media.
Liu, B. (2012). Sentiment analysis and opinion mining. Synthesis lectures on human language technologies, 5(1), 1–167.
Wilie, B., Vincentio, K. R., Winata, G. I., Cahyawijaya, S., Li, X., Fung, P., & Purwarianti, A. (2020). IndoNLU: Benchmark and resources for evaluating Indonesian natural language understanding. arXiv preprint arXiv:2009.05387.
Saputri, G. R., Sari, Y. A., & Nugroho, R. A. (2021). IndoBERT for Indonesian Sentiment Analysis: A Case Study on Twitter Data. Journal of Information Systems Engineering and Business Intelligence, 7(2), 107-115.
Kouloumpis, E., Wilson, T., & Moore, J. (2011). Twitter sentiment analysis: The good the bad and the OMG!. Proceedings of the Fifth International Conference on Weblogs and Social Media.
Setiawan, A., Putra, A., & Rachmawati, D. (2020). Sentiment Analysis on Twitter Using BERT-based Indonesian Language Model. 2020 International Conference on Data Science and Its Applications (ICoDSA), 1–5.
Hutto, C.J. & Gilbert, E.E. (2014). VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text. Proceedings of the International AAAI Conference on Weblogs and Social Media, 8(1), 216-225
Raharjo, P., Irawan, B., & Nugroho, H. A. (2021). Normalization of Indonesian social media texts using hybrid approach. Journal of Information Technology & Computer Science, 9(2), 97–104.
Wolf, T., et al. (2020). Transformers: State-of-the-art Natural Language Processing. EMNLP.
Liu, Y., Melas-Kyriazi, L., Goyal, N., & Lewis, M. (2020). Multilingual Denoising Pre-training for Neural Machine Translation. Transactions of the Association for Computational Linguistics, 8, 726–742.
Ahuja, A., & Dubey, P. (2019). Pre-processing Twitter data for sentiment analysis. International Journal of Scientific & Technology Research, 8(11), 3701–3703.
Wijaya, D. T., & Saputra, D. (2020). Analisis sentimen opini masyarakat Indonesia terhadap pemerintah selama pandemi COVID-19 menggunakan metode LSTM. Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi), 4(3), 458–464. https://doi.org/10.29207/resti.v4i3.2060
Oktavia, A. D., & Wulandari, A. (2021). Penerapan Algoritma BERT dalam Klasifikasi Sentimen pada Twitter dengan Bahasa Indonesia. Jurnal Teknologi dan Sistem Komputer, 9(4), 377–384. https://doi.org/10.14710/jtsiskom.9.4.2021.377-384
Baziotis, C., Pelekis, N., & Doulkeridis, C. (2017). DataStories at SemEval-2017 Task 4: Deep LSTM with Attention for Message-level and Topic-based Sentiment Analysis. Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017), 747–754. https://doi.org/10.18653/v1/S17-2127
Soetikno, D., & Susanto, D. H. (2022). Analisis sentimen masyarakat terhadap kebijakan publik di Twitter menggunakan BERT. Jurnal Ilmu Komputer dan Aplikasinya, 21(1), 12–21.
Hidayatullah, M. F., & Nuryani, L. (2021). Analisis Sentimen pada Twitter menggunakan Algoritma Support Vector Machine dan Naive Bayes. Jurnal Teknologi dan Sistem Komputer, 9(1), 14–20. https://doi.org/10.14710/jtsiskom.9.1.2021.14-20
Fatmawati, D., & Nurhasanah, L. (2022). Penerapan Preprocessing dan Tokenisasi dalam Analisis Sentimen Data Media Sosial. Jurnal Sistem Informasi dan Komputer, 6(2), 115–122.
