Table 11 Comparison to benchmark results
From: A neural network multi-task learning approach to biomedical named entity recognition
Corpus | Benchmark | Ours | Matching criteria |
|---|---|---|---|
AnatEM | 91.61 | 88.55 | Right boundary match |
BC2GM | 87.21 | 84.41 | Alternative boundaries |
BC4CHEMD | 87.39 | 82.32 | Exact |
BC5CDR | 86.76 | 83.87 | Exact |
JNLPBA | 72.55 | 68.95 | Exact |
Linnaeus | 95.68* | 79.33 | Exact |
NCBI-Disease | 82.9 | 77.82 | Exact |