ANALISIS POLARISASI OPINI BBM MENGGUNAKAN GCN DAN PMI
DOI:
https://doi.org/10.30656/8pz6gc44Abstract
This study analyzes the polarization of public opinion regarding the removal of fuel subsidies in Indonesia using a graph-based sentiment classification approach that integrates Pointwise Mutual Information (PMI) and Graph Convolutional Networks (GCN). Sentiment analysis of social media data presents challenges due to the short and noisy nature of texts, which often limits the performance of conventional classification methods in capturing contextual and semantic relationships. Additionally, many existing approaches primarily rely on vector-based text representations and do not explicitly capture semantic relationships between words within a graph structure. To address this limitation, this study proposes a graph-based representation that incorporates PMI to strengthen semantic relationships between words in the construction of a GCN model. A dataset consisting of 557 tweets related to the fuel subsidy issue was collected and preprocessed before being transformed into a graph representation. In this graph representation, document-word relationships were weighted using TF-IDF, while word-word relationships were weighted using PMI to capture co-occurrence-based semantic associations. Two experimental scenarios were evaluated: a baseline GCN model without PMI and a PMI-enhanced GCN model. Both models were trained using a 70:15:15 train-validation-test split with identical hyperparameter settings. The experimental results show that the baseline GCN achieved an accuracy of 65%, while the integration of PMI improved the accuracy to 73%. These findings indicate that incorporating PMI enriches the semantic relationships within the graph representation and improves the effectiveness of GCN in sentiment classification tasks. The proposed approach contributes to the development of graph-based sentiment analysis methods and provides insights into public opinion polarization related to government policy issues on social media.
Keywords— Graph Convolutional Networks, Pointwise Mutual Information, Sentiment Analysis, Public Opinion, Social Media, Fuel Subsidy Policy.
References
[1] Zulham, A. Lubis, D. Priyono, Fauzan, S. Julina, and A. Deryansyah, “ANALISIS FRAMING MEDIA DALAM BERITA KONTROVERSIAL: STUDI KASUS PADA KASUS-KASUS POLITIK ATAU SOSIAL,” Jurnal Review Pendidikan dan Pengajaran, vol. 7, no. 3, pp. 9118–9126, 2024, doi: 10.31004/jrpp.v7i3.30952.
[2] H. Sujadi, S. Fajar, and C. Roni, “ANALISIS SENTIMEN PENGGUNA MEDIA SOSIAL TWITTER TERHADAP WABAH COVID-19 DENGAN METODE NAIVE BAYES CLASSIFIER DAN SUPPORT VECTOR MACHINE,” INFOTECH journal, vol. 8, no. 1, pp. 22–27, Jun. 2022, doi: 10.31949/infotech.v8i1.1883.
[3] Y. Romadhoni and K. Holle, “Analisis Sentimen Terhadap PERMENDIKBUD No.30 pada Media Sosial Twitter Menggunakan Metode Naive Bayes dan LSTM,” Jurnal Informatika: Jurnal Pengembangan IT (JPIT), vol. 7, no. 2, pp. 118–124, Apr. 2022.
[4] R. Ardiansyah, N. Inayah, and M. Liebenlito, “Analisis Sentimen Publik Terhadap Penangkapan Ikan Ilegal oleh Kapal Vietnam di Twitter Menggunakan Metode GRAFT,” The Indonesian Journal of Computer Science, vol. 14, no. 2, Apr. 2025, doi: 10.33022/ijcs.v14i2.4650.
[5] S. Muhsini, “ANALYSIS AND IMPLEMENTATION OF CROSS-LINGUAL SEMANTIC SIMILARITY WORDS USING POINTWISE MUTUAL INFORMATION METHOD,” Jurnal Penelitian Pendidikan, vol. 18, no. 1, pp. 18–24, 2018, doi: 10.17509/jpp.v18i1.11056.
[6] K. Wang, Y. Ding, and S. C. Han, “Graph neural networks for text classification: a survey,” Artif. Intell. Rev., vol. 57, no. 8, pp. 1–38, Jul. 2024, doi: 10.1007/s10462-024-10808-0.
[7] M. Ameripour and V. Seydi, “Sentiment Analysis with Graph Convolutional Networks using directed PMI,” Research Square preprint, Jun. 2024, doi: 10.21203/rs.3.rs-4564482/v1.
[8] A. Khosravi, Z. Rahmati, and A. Vefghi, “Relational Graph Convolutional Networks for Sentiment Analysis,” arXiv preprint arXiv:2404.13079, Apr. 2024, doi: 10.48550/arXiv.2404.13079.
[9] R. Faturahman, N. Azahra, A. Triana Perdika, F. Mursidul Akmal, N. Rizqia Kamila, and P. Khoirotun, “Polarisasi Sebagai Bentuk Efek Negatif Dalam Pembentukan Opini Publik Pada Pemilihan Presiden 2019,” INTERAKSI PERADABAN: Jurnal Komunikasi dan Penyiaran Islam, vol. 4, no. 1, pp. 99–115, 2024, doi: 10.15408/interaksi.v4i1.38849.
[10] S. Hidayat Daeli et al., “POLARISASI POLITIK DI MEDIA SOSIAL : TELAAH FILSAFAT PERSATUAN INDONESIA DI ERA DIGITAL,” Triwikrama: Jurnal Ilmu Sosial, vol. 5, no. 10, pp. 1–5, 2023.
[11] D. Arisandi, T. Sutrisno, and I. Kurniawan, “KLASIFIKASI OPINI MASYARAKAT DI TWITTER TENTANG KEBOCORAN DATA YANG TERJADI DI INDONESIA MENGGUNAKAN ALGORITMA SVM,” Jurnal Informatika Kaputama (JIK), vol. 7, no. 1, pp. 84–90, Jan. 2023, doi: 10.59697/jik.v7i1.10.
[12] S. Maula Chamzah, M. Lestandy, N. Kasan, and A. Nugraha, “Penerapan Synthetic Minority Oversampling Technique (SMOTE) untuk Imbalance Class pada Data Text Menggunakan KNN,” Syntax: Jurnal Informatika, vol. 11, no. 02, pp. 56–67, 2022.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Chalifa Chazar, Dea Amelia Azzahra, Asep Nana Hermana, Marisa Premitasari
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
- This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
-
Author(s)' Warranties
The author warrants that the article is original, written by stated author(s), has not been published before, contains no unlawful statements, does not infringe the rights of others, is subject to copyright that is vested exclusively in the author and free of any third party rights, and that any necessary written permissions to quote from other sources have been obtained by the author(s).
- Information
- Notice about change in the copyright policy of the journal 'Jurnal Sistem Informasi (JSiI)' : "From Vol 1, onwards the copyright of the article published in the journal 'Jurnal Sistem Informasi' will be retained by the author"
