Table 2.
Isolatedness metric for several sets of papers
n | BERT (%) | TF-IDF (%) | |
---|---|---|---|
COVID-19 | 132,802 | 80.6 | 76.2 |
HIV/AIDS | 308,077 | 63.9 | 62.3 |
Influenza | 90,575 | 57.9 | 64.1 |
Meta-analysis | 145,358 | 52.6 | 38.5 |
Virology | 112,807 | 47.7 | 39.1 |
Ophthalmology | 144,411 | 47.7 | 43.6 |
Fraction of k nearest neighbors of papers from each corpus that also belong to the same corpus (see experimental procedures). The first four rows show corpora selected based on the abstract text; the last two, based on the journal name.