Pemetaan Penyebaran Topik Percakapan Pengguna KRL (Commuter Line) Pada Media Sosial Twitter Menggunakan SNA Dan Bertopic

Marsyanda Nur Zahra (Universitas Negeri Surabaya)
Monica Cinthya (Universitas Negeri Surabaya)

Jurnal Ilmu Ekonomi, Pendidikan dan Teknik (IDENTIK)
Vol. 3 No. 6 (2026): IDENTIK - November

KRL (Commuter Line) is one of the main public transportation services used by the community, particularly in the Greater Jakarta area. The high level of activity among KRL users has generated various conversations on social media X (Twitter) concerning schedules, routes, operational disruptions, complaints, and service facilities. The large volume of unstructured conversations makes it difficult to manually identify the topics discussed and the relationships among them. This study aims to map the topics of KRL Commuter Line user conversations on social media X using BERTopic and analyze the relationships among topics using Social Network Analysis (SNA). The data were collected through web scraping from social media X using the keywords "KRL", "Commuter Line", and "@CommuterLine" during the period from March to May 2026, and were subsequently processed through a preprocessing stage. BERTopic modeling consisted of sentence embedding using SentenceTransformer, dimensionality reduction using UMAP, clustering using HDBSCAN, and topic representation using C-TF IDF. The modeling results identified ten main topics covering schedules and delays, routes and travel directions, operational disruptions, unresolved user complaints, women's carriages, and priority seating, with a Topic Coherence score of 0.4387 and a Topic Diversity score of 0.7897. SNA was then used to map the relationships among topics based on shared keywords, forming a network with 9 nodes and 12 edges, an Average Degree of 1.333, an Average Weighted Degree of 2.778, and three communities with a modularity value of 0.337. The network evaluation showed that Topic 3 (Routes and Stations) and Topic 4 (Unresolved Complaints) had the highest degree value of 4, while Topic 4 had the highest betweenness centrality and served as a bridge between groups. The network of KRL user conversation topics was divided into three communities: schedules and routes, operational disruptions and complaints, and priority facilities, which were interconnected through Topic 3 and Topic 4. Therefore, the combination of BERTopic and SNA provides an overview of KRL user conversation topics and the relationships among topics on social media X.