Table 9 Pearson’s correlation between general-domain datasets/Bio-SimVerb and Bio-SimLex scores and the NER tasks evaluated on general-domain representation models benchmarked in SimVerb and SimLex. None of the scores are statistically significant
From: Bio-SimVerb and Bio-SimLex: wide-coverage evaluation sets of word similarity in biomedicine
| Â | BC4CHEMD | BC2GM | AnatEM | JNLPBA |
|---|---|---|---|---|
SimVerb | -0.31 | -0.09 | -0.41 | -0.12 |
SimLex | -0.36 | -0.20 | -0.49 | -0.19 |
Bio-SimVerb | -0.38 | -0.18 | -0.47 | -0.22 |
Bio-SimLex | 0.00 | 0.23 | -0.09 | 0.18 |