المفاهيم الأساسية
Real-world stance detection approaches need to consider temporal factors for accurate COVID-19 vaccination stance classification.
الملخص
1. Abstract:
Importance of vaccination in controlling COVID-19 transmission.
Evolution of attitudes towards vaccination on social media.
Impact of temporal concept drift on stance detection.
2. Introduction:
Promoting vaccination as a strategy to curb the virus spread.
Varied attitudes towards COVID-19 vaccination.
Growing interest in using machine learning for stance detection.
3. Dataset and Research Questions:
Evaluation of datasets with different timeframes and languages.
Research questions on temporal concept drift, model performance, and domain adaptation.
4. Experimental Setup:
Use of monolingual and multilingual datasets.
Evaluation of transformer-based PLMs with chronological and random splits.
5. Analysis:
Text Similarity:
IoU and DICE coefficients used to measure similarity between training and test sets.
Topics Drift:
BERTopic analysis to examine topic distribution over time.
Error Analysis:
Model behaviors observed with random vs. chronological splits.
6. Discussion:
Impact of split strategies on model performance across datasets.
Effectiveness of domain-adapted PLMs in addressing temporal concept drift.
7. Conclusion:
Chronological splits reduce accuracy in stance classification.
Importance of considering temporal factors in real-world stance detection approaches.
الإحصائيات
"Time Tweets Labels Language Cotfas et al. (2021) Nov 2020 ∼Dec 2020 2,792 in favour, against, neutral en Poddar et al. (2022a) Jan 2020 ∼March 2021 1,700 in favour, against, neutral en Mu et al. (2023b) Nov 2020 ∼April 2022 3,101 pro, anti, hesitancy, irrelevant en Chen et al. (2022) Jan 2020 ∼March 2021 17,934 pos, neg, neutral, off-topic fr, de, en Di Giovanni et al. (2022) Nov 2020 ∼June 2021"
اقتباسات
"Promoting vaccination has been statistically recognised as a vital tactic in curbing the spread of the COVID-19 virus."
"It is crucial for policymakers to have a comprehensive understanding of the public’s stance towards vaccination on a large scale."