Deep Learning-Based Sentiment and Stance Analysis of Tweets About Vaccination

Küçük D., ARICI N.

International Journal on Semantic Web and Information Systems, cilt.19, sa.1, 2023 (SCI-Expanded) identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 19 Sayı: 1
  • Basım Tarihi: 2023
  • Doi Numarası: 10.4018/ijswis.333865
  • Dergi Adı: International Journal on Semantic Web and Information Systems
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Communication Abstracts, Compendex, INSPEC, Library and Information Science Abstracts, Metadex, Civil Engineering Abstracts
  • Anahtar Kelimeler: Affective Computing, BERT, ChatGPT, Deep Learning, Health Informatics, Pre-trained Transformers, Public Health, Social Media Analysis
  • Gazi Üniversitesi Adresli: Evet

Özet

Sentiment analysis and stance detection are interrelated problems of affective computing, and their outputs commonly complement each other. The focus of this article is to determine sentiments and stances of Twitter users about vaccination. A tweet dataset on COVID-19 vaccination is compiled and jointly annotated with sentiment and stance. This deep learning approach employs BERT, which is a model based on pre-trained transformers. The generative deep learning model, ChatGPT, is also used for stance and sentiment analysis on the dataset. ChatGPT achieves the best performance for stance detection, while BERT is the best performer for sentiment analysis. This study is the first one to observe stance and sentiment detection performance of ChatGPT on health-related tweets. This article also includes a full-fledged system proposal based on automatic sentiment and stance analysis. COVID-19 pandemic is an impactful global public health phenomenon, and hence, joint extraction of sentiments and stances from health-related tweets can profoundly contribute to health-related decision-making processes.