An Intelligent COVID-19-Related Arabic Text Detection Framework Based on Transfer Learning Using Context Representation

Table 1

A summary of the latest studies related to COVID-19 Arabic text detection (ATD).

Ref	Objective	Representation methods	Dataset	Evaluation metrics	Pros	Cons
[20]	Created a dataset called “AraCOVID19-MFH” for multilabel	BERT	AraCOVID19MFH dataset	F1-sarcastic with data augmentation 86 but without was 71	Good performance	Data not available
[21]	Built and released AraCOVID-19-SSD1	DL	Dataset with 5,162 tweets	All	Sarcasm and sentiment detection	Data not available
[13]	Aimed to identify hate speech related to the COVID-19 pandemic	DL and topic modeling	Twitter data	All	Twitter data in the Arabic region	—
[22]	Introduced ArCOV19-Rumors, an Arabic COVID-19 Twitter dataset for misinformation detection	—	Create COVID-19 Twitter dataset	All	Available	They do not consider different topics
[14]	Detecting inauthentic news about COVID-19 in Arabic tweets was addressed	—	Collected nearly 7 million Arabic tweets	—	—	Data not available
[7]	Proposed a hybrid model for detecting COVID-19-	Concatenate LSTM and parallel CNN	COVID-19 Twitter dataset	All		Need more time
[12]	Proposed a new approach based on ensemble techniques for	Ensemble techniques	COVID-19 Twitter dataset	All	Detecting and tracking COVID-19 rumors	Need more time
[19]	Collected data on three topics	ML and DL	A dataset called ArCOVID-19 Vac	Accuracy equal to 86.4, 75.4, and 82.2	Manually annotated	Their data more imbalance