|
KMID : 0365220220590010001
|
|
Korean Journal of Public Health
2022 Volume.59 No. 1 p.1 ~ p.11
|
|
|
Keyword Analysis in Korean Articles Related to COVID-19 Using Topic Modeling
|
|
Jeon Eun-Su
Oh Seung-Hoon
Cho Yeong-Mok
|
|
|
Abstract
|
|
|
Objectives: There was much research that assessed public health policies regarding the COVID-19 pandemic or analyzed media reports on other issues, but statistical analysis on COVID-19 related news reports were scarce. Thus, this study aims to apply LDA (Latent Dirichlet Allocation; topic modeling) method to news reports, and track the effects and interest related to pandemic policies.
Methods: 182,922 news articles from 11 main daily newspapers in Korea during 2020.10.12. ~ 2021.07.19. were used for morpheme analysis and topic classification using Mallet LDA method. 22 topics were decided by coherence score, then visualized by PCA (Principal Component Analysis) and compared with each other. Report statistics were also compared to the timeline of COVID-19 measures, particularly social distancing and vaccination policies.
Results: By comparing confirmed cases with the trend of keywords 'confirmed case status', 'social distancing guidelines', and 'vaccine inoculation', changing patterns in topics over time could be observed, and the cause could be analyzed. Especially, there was a significant difference between the third and fourth wave of the pandemic. At least in the aspect of media focus, the effectiveness of social distancing policies decreased in the fourth wave. Also, there was a significant media interest with the vaccine inoculation started.
Conclusion: Concluding from the media focus level, vaccination policies more influence rather than social distancing policies now.
|
|
|
KEYWORD
|
|
|
COVID-19, media report, LDA, topic classification
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
|
|
|