Skip to main content

Table 2 Performances of single-task models

From: CollaboNet: collaboration of deep neural networks for biomedical named entity recognition

Model Habibi et al. (2017) STM Wang et al. (2018) STM Our STM
Dataset Precision Recall F1 Score Precision Recall F1 Score Precision Recall F1 Score
NCBI-disease 85.31 83.58 84.44 84.95 82.92 83.92 83.95 85.45 84.69 (±0.54)
JNLPBA 74.83 79.82 77.25 69.60 74.95 72.17 72.51 82.98 77.39 (±0.24)
BC5CDR-chem 92.57 88.77 90.63 *93.05 *86.87 *89.85 94.02 91.50 92.74 (±0.47)
BC5CDR-disease 84.19 82.79 83.49 *84.09 *81.32 *82.68 82.98 82.25 82.61 (±0.25)
BC4CHEMD 87.83 85.45 86.62 90.53 87.04 88.75 90.50 85.96 88.19 (±0.23)
BC2GM 77.50 78.13 77.82 81.11 78.91 80.00 79.70 77.47 78.56 (±0.38)
Macro Average 83.71 83.09 83.38 83.89 82.00 82.90 83.94 84.27 84.03
  1. Our STM achieved the best performance on 3 datasets among 6. Scores in the asterisked (*) cells are obtained in the experiments that we conducted; these scores are not reported in the original papers. The best scores from these experiments are in bold