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

Table 4

COVID-19 data distribution for fine-grained content of multiclass classification scenario.

S. no.	Category fine-grained content	No. of Arabic documents	Training/validation	Testing
1	Info-news	5,217	4,160	1,057
2	Celebrity	1,396	1,120	276
3	Plan	860	688	172
4	Requests	172	132	40
5	Rumors	118	94	24
6	Advice	94	77	17
7	Restrictions	24	20	4
8	Personal	1,427	1,153	274
9	Unrelated	449	359	90
10	Others	229	184	45
Total		9988	9,986	7,987