A lot of people concerns about radiation exposure by Fukushima nuclear accident. But no method that we effectively understand people's anxiety because we grasped people's needs about radiation exposure with individualism (ex. Telephone, questionnaire). In this paper, we suppose that it is possible to efficiently collect more people's needs by utilizing twitter, which is one of Social Network Service, easier than before. Firstly, we search tweet including words of (“radiation” and “exposure”), normalize unicode, process pointless words (ex. URL, html tags, @NAME). Secondly, we extract tweets with anxious words dictionary that there are low dependency on texture and annotate 999 tweets to Anxious or Nonanxious. Thirdly, we extract tweets by sentiment analysis. Finally, we analysis tweets by utilizing KHCoder. In this dataset, about 40 % opinion data are anxiety about medical exposure. We infer that people are concerned about radiation exposure related to medical treatment, which is relatively close to everyday life. In this paper, we developed the system, which from collecting anxious opinion about health effects due to radiation exposure by using Twitter to doing dataset tendency analysis on about 60% Precision and Recall, by utilizing anxiety dictionary and sentiment analysis.
|Translated title of the contribution||Development of method using sentiment analysis for anxiety opinion of radiation exposure by social big data|
|Number of pages||8|
|Journal||Japanese Journal of Health Physics|
|Publication status||Published - 2020|
All Science Journal Classification (ASJC) codes
- Radiology Nuclear Medicine and imaging
- Health, Toxicology and Mutagenesis