Table 8 A qualitative comparison of related research studies according to several features such as approach, analysis scope, sentiment analysis tool and shifts, data collection, and duration of analysis, etc.
Research Study | Approach | scope | Text/Image Based | Sentiment analysis tool | Sentiment shift analysis | Dataset / Data Source | Model used | Duration of analysis |
|---|---|---|---|---|---|---|---|---|
Boonyarat, Liew, Chang (2024) | BERT model for suicidal ideation detection | Thai-specific during the pandemic | Text | BERT model based | No | Thai Social media content | BERT model | Pandemic specific |
Zeng, Sun, and Li (2023) | Multimodal negative sentiment recognition | Public health emergencies | Text and visual data | GCN-based | Negative sentiment shifts only | Text and image from public opinion | GCN and ensemble learning | Public health emergencies during the pandemic |
A. Chen et al. (2022) | Analysis of social representations during COVID-19 | China-focused | Text | Lexicon based | Dynamics of social representation over time | Chinese social media | Social representation model | 2019-2020 |
J. Gao et al. (2022) | Comparative analysis of WFH challenges | Comparative analysis of WFH sentiment between platforms | Text | Lexicon based sentiment analysis | No | Weibo, Twitter | N/A | During early-to-mid pandemic |
Sitaula et al. (2021) | Deep learning based sentiment analysis | Sentiment analysis on nepali tweets during the pandemic | Text | Bi-LSTM, CNN | No | Nepali twitter | Bi-LSTM, CNN | Pandemic specific period |
Sharma et al. (2020) | Analysis of free-form text to assess student experiences | Analysis of student-written free-form text to assess COVID-19 impacts | Text | NLTK-based sentiment classifier | Assessed sentiment changes over time | Free-form text from students | Supervised ML, unsupervised clustering | Focused on pandemic’s impact on students |
Ni et al. (2020) | Survey-based cross-sectional study | Mental health and social media use analysis | N/A | N/A | No | Survey data from wuhan | N/A | Early pandemic phase |
GarcÃa-DÃaz et al. (2018) | Text-based sentiment analysis on infectious disease-related tweets | Focused on infectious diseases in South America | Text | N/A | No | Tweets related to infectious diseases from South American users | N/A | Not pandemic specific, focused on multiple infectious diseases |
Nagel et al. (2013) | Text-based analysis of the link between real-world events and tweets about influenza and pertussis | Focused on understanding the relationship between real-world outbreaks and sentiment changes | Text | N/A | Correlation between realspace events and online sentiment changes | Tweets related to influenza and pertussis outbreaks | N/A | Focused on real-world events during specific outbreaks (Influenza, Pertussis) |
Our Study | Bimodal longitudinal sentiment analysis | Global, focused on general sentiment changes | Bimodal (Text and Image) | Custom sentiment classifier | Longitudinal approach across multiple phases | Large multimodal dataset (text+image) from Twitter and Instagram | CNN, LSTM, Hybrid deep learning model | Pre-pandemic, Post-pandemic |
