Table 4 Evaluation of models trained on combined NCBI and BC5CDR data set
From: We are not ready yet: limitations of state-of-the-art disease named entity recognizers
Algorithm | Test set | Precision[%] | Recall[%] | F1-Score[%] |
|---|---|---|---|---|
TaggerOne | NCBI | 81.86 | 80.23 | 81.04 |
BC5CDR | 79.61 | 77.69 | 78.64 | |
BioBERT | NCBI | 85.19 | 88.74 | 86.93 |
BC5CDR | 82.07 | 85.21 | 83.61 |