Biomedical named entity recognition with the combined feature attention and fully-shared multi-task learning

BMC Bioinformatics

Table 2 Performance of different single-task models

Model	BC2GM	JNLPBA	BC5CDR-Disease	NCBI-disease	Linnaeus	Species-800	BC5CDR-Chemical
BioBERT	84.72	77.49	87.15	89.71	88.24	74.06	93.47
BioBERT (ours)	84.62	77.41	86.89	89.18	88.06	74.88	93.44
BioKMNER
POS	84.74	77.06	–	89.47	88.44	75.45	93.73
Syn	84.76	77.17	–	89.27	88.68	75.65	93.74
Dep	84.65	77.32	–	89.24	88.57	75.81	93.78
BioBERT-TWA (ours)
POS	84.83	77.81	86.91	89.41	88.17	75.03	93.52
Syn	84.96	78.02	87.03	89.36	88.42	75.24	93.68
Dep	84.85	78.23	87.12	89.52	88.36	75.14	93.81
BioBERT-CFA (ours)
POS	85.06	78.21	87.45	89.98	88.38	75.62	93.89
Syn	85.36	78.36	87.49	89.91	88.75	75.83	94.05
Dep	85.28	78.47	87.56	89.94	88.66	75.64	94.09

ISSN: 1471-2105