Skip to main content

Table 5 Contrast experiment results between different pre-trained language models and different external features

From: External features enriched model for biomedical question answering

Model 6b Factoid QA 7b Factoid QA 8b Factoid QA
SAcc LAcc MRR SAcc LAcc MRR SAcc LAcc MRR
SciBERT 0.3688 0.5974 0.4544 0.4203 0.6051 0.4919 0.3793 0.5737 0.4496
SciBERT+POS+NER 0.3967 0.5959 0.4766 0.4253 0.5901 0.4900 0.3874 0.5523 0.4499
Baseline (BioBERT) 0.3973 0.6217 0.4838 0.4318 0.6164 0.5007 0.3848 0.5585 0.4492
BioBERT+BioNER+FF 0.4380 0.6075 0.5078 0.4355 0.6113 0.5012 0.3982 0.5875 0.4692
BioBERT+POS+BioNER+FF 0.4171 0.6263 0.5011 0.4419 0.6211 0.5106 0.4032 0.5858 0.4689
Our model (BioBERT+POS+NER+FF) 0.4517 0.6294 0.5197 0.4444 0.6419 0.5165 0.3937 0.6098 0.4688
  1. Bold values represent the highest results