Table 2 The statistic information of different CovSI datasets
From: Multi-probe attention neural network for COVID-19 semantic indexing
Type | Training set | Development set | Test set |
|---|---|---|---|
#Articles | 71,207 | 8,000 | 8,000 |
#MeSH term types | 17,758 | 9,035 | 8,991 |
#Total terms | 945,462 | 106,088 | 110,412 |
#Average terms per article | 13.28 | 13.26 | 13.80 |