Skip to main content
. 2018 Feb 5;19:33. doi: 10.1186/s12859-018-2039-z

Table 8.

Pearson’s correlation between word-similarity/Bio-SimVerb and Bio-SimLex scores and the NER tasks evaluated on biomedical representation models trained with different window sizes

BC4CHEMD BC2GM AnatEM JNLPBA
UMN-rel -0.78a -0.56 -0.78a -0.46
UMN-sim -0.73 -0.57a -0.81 -0.42a
MayoSRS -0.78 -0.69 -0.54a -0.47a
Bio-SimVerb 0.63 0.36 0.42 0.40
Bio-SimLex 0.83 a 0.66 0.92 a 0.59

Bold: best scores

aStatistically significant