Clustering Public Discourse on Indonesian Coal-Fired Power Plants: A K-Means Approach for Social Media and News

This research explores public discourse surrounding Indonesian coal-fired power plants (CFPPs) by utilizing a computational approach to analyze social media and news content through K-Means clustering. CFPPs play a central role in Indonesia’s energy production while simultaneously raising concerns regarding environmental pollution and public health risks. Conversations around CFPPs frequently appear on digital platforms such as Twitter as well as in news reporting. However, the unstructured and fragmented nature of these discussions presents challenges for systematic analysis. This study addresses that challenge by analyzing a dataset of 7,910 Indonesian-language documents collected between January 2023 and December 2024, comprising 6,895 tweets and 1,015 news articles. A multi-stage methodology was applied, incorporating text preprocessing, bigram generation, Word2Vec embeddings, and K-Means clustering to identify thematic patterns in the discourse. The Elbow Method and silhouette coefficient guided the selection of three clusters, with the final clustering yielding a silhouette score of 0.638, indicating coherent group formation. Further analysis identified that Cluster 1 predominantly captures discussions related to environmental concerns, specifically air pollution and its local impacts. Cluster 2 focuses on technical aspects of electricity generation, including operational details of CFPPs and related energy infrastructure. Cluster 3 highlights broader fossil fuel-related topics, with discussions on gas-powered plants and regional energy developments. Duplicate detection revealed 606 repeated documents across the dataset, indicating patterns of content replication within the discourse. Visualizations, including bigram frequency charts and treemaps, were used to illustrate the thematic composition of each cluster. These findings contribute to a systematic understanding of the narratives shaping public discussions on CFPPs in Indonesia.