Temporal and Sentiment Analysis of Alcohol-Related Tweets in Thailand During 2023: Patterns and Trends in Online Discourse
Issued Date
2026-01-01
Resource Type
eISSN
24654418
Scopus ID
2-s2.0-105017679328
Journal Title
Journal of Population and Social Studies
Volume
34
Start Page
592
End Page
614
Rights Holder(s)
SCOPUS
Bibliographic Citation
Journal of Population and Social Studies Vol.34 (2026) , 592-614
Suggested Citation
Lamy F.R., Paek S.C., Meemon N. Temporal and Sentiment Analysis of Alcohol-Related Tweets in Thailand During 2023: Patterns and Trends in Online Discourse. Journal of Population and Social Studies Vol.34 (2026) , 592-614. 614. doi:10.25133/JPSSv342026.030 Retrieved from: https://repository.li.mahidol.ac.th/handle/123456789/114745
Title
Temporal and Sentiment Analysis of Alcohol-Related Tweets in Thailand During 2023: Patterns and Trends in Online Discourse
Author's Affiliation
Corresponding Author(s)
Other Contributor(s)
Abstract
There is a paucity of research concerning the content and temporalities of alcohol-related social media texts posted in the Thai language. A total of 12,065,726 tweets were collected between January 1, 2023, and May 23, 2023, based on thirteen alcohol-related . Three native Thai speakers manually coded 15,000 random tweets to explore the type, sentiment, and content of collected tweets. “Personal communications” represented 49.1% of the sample, 15.3% were coded as “Pornography-related,” and 35.6% as “Irrelevant.” Among the personal communication tweets, 81.1% were coded as Neutral, 4.9% as Positive, and 14.0% as Negative. Despite a higher volume of negative tweets, only one prevention-oriented tweet was found during the qualitative content analysis. The coded tweets were further used to train supervised machine learning algorithms to identify posts labelled as positive, neutral, and negative within the whole dataset. Temporal heatmaps of positive, neutral, and negative personal communication tweets were then generated. Negative tweets were more likely to be posted on Sunday evening (from 17:00 to midnight) and Monday early afternoon (13:00 to 15:00), while positive tweets were frequently posted on the evenings (after 20:00), especially Monday. Our results can be used to disseminate alcohol-related health prevention messages at the time and day(s) of the week when such messages would be most read on X.
