COVID-19 Fake News and Misinformation Detection using Transformer Learning
3rd International Conference on Education, Knowledge and Information Management, ICEKIM 2022
; : 965-968, 2022.
Article
in English
| Scopus | ID: covidwho-2255893
ABSTRACT
As COVID-19 spreads globally and generates an unprecedented pandemic, COVID-19 fake news is born and quickly disseminated on the internet. Misinformation and disinformation of COVID-19 can distort public perception of the virus and have a serious negative influence on society. To increase vaccine coverage rates and achieve herd immunity, eliminating fake news becomes an urgent need worldwide. Our research aims at using the Transformer model to implement COVID-19 fake news detection. We use the dataset of COVID-19 fake news, extract features through the embedding method of one hot representation, and construct the transformer model to implement text classification on the binary problem. Then we analyze results through loss curve and confusion matrix and show performance parameters, including accuracy, AUC score, and F1 score. We conclude that the model can achieve an accuracy of 72% for COVID-19 fake news detection. This research provides insight for transformer learning dealing with fake news detection of COVID-19. © 2022 IEEE.
Full text:
Available
Collection:
Databases of international organizations
Database:
Scopus
Language:
English
Journal:
3rd International Conference on Education, Knowledge and Information Management, ICEKIM 2022
Year:
2022
Document Type:
Article
Similar
MEDLINE
...
LILACS
LIS