<![CDATA[Online media has become the primary source of public health news due to its ease of access and wide dissemination of information. However, the abundance of information makes it difficult to identify key issues and map public opinion trends, resulting in a mismatch between information needs and the necessary policy responses. This study aims to automatically model public health news topics in online media to reveal latent structures and quantify media attention to various health issues. A total of 8,799 public health news articles from the national news portals Detik and Kompas were analyzed using text pre-processing, numerical corpus formation using the Bag of Words method, topic modeling using Latent Dirichlet Allocation, and evaluation using the C_v coherence metric. The test results showed that the optimal model was obtained for three main topics with a coherence C_v value of 0.61. The distribution of media attention to these three topics can be quantified as follows: free health services at 46.4%, community lifestyle and nutrition at 35.3%, and health insurance policies and systems such as BPJS and JKN at 18.4%, which are visualized with clear topic separation and stable semantic relationships using pyLDAvis. ]]>
Copyrights © 2026
