Title: Sentiment analysis and counselling for COVID-19 pandemic based on social media

Authors: Ha-Young Lee; Ok-Ran Jeong

Addresses: School of Computing, Gachon University, Seongnam-si, Gyeonggi-do, 13120, Korea ' School of Computing, Gachon University, Seongnam-si, Gyeonggi-do, 13120, Korea

Abstract: As COVID-19 emerged and prolonged, various changes have occurred in our lives. For example, as restrictions on daily life are lengthening, the number of people complaining of depression is increasing. In this paper, we conduct a sentiment analysis by modelling public emotions and issues through social media. Text data written on Twitter is collected by dividing it into the early and late stages of COVID-19, and emotional analysis is performed to reclassify it into positive and negative tweets. Therefore, subject modelling is performed with a total of four datasets to review the results and evaluate the modelling results. Furthermore, topic modelling results are visualised using dimensional reduction, and public opinions on COVID-19 are intuitively confirmed by generating representative words consisting of each topic in the word cloud. Additionally, we implement a COVID-chatbot that provides a question-and-answer service on COVID-19 and verifies the performance in our experiments.

Keywords: social media analysis; sentiment analysis; topic modelling; COVID-chatbot; Google BERT; Microsoft DialoGPT.

DOI: 10.1504/IJWGS.2023.129327

International Journal of Web and Grid Services, 2023 Vol.19 No.1, pp.34 - 57

Received: 04 May 2022
Received in revised form: 31 Aug 2022
Accepted: 01 Sep 2022
Published online: 06 Mar 2023 *