SBLC: a hybrid model for disease named entity recognition based on semantic bidirectional LSTMs and conditional random fields

Table 8 The performance comparison of our SBLC model with the baseline methods on the same NCBI test dataset

Methods	Precision	Recall	F1
Dictionary look-up [2]	0.213	0.718	0.316
cTAKES (version 4.0) [15]	0.476	0.541	0.506
MetaMap (semantic type filtering) [14]	0.495	0.679	0.541
MetaMap (MEDIC filtering) [14]	0.510	0.702	0.559
Inference method [2]	0.597	0.731	0.637
CRF + CMT [34]	0.795	0.683	0.735
CRF + MeSH [34]	0.855	0.660	0.746
CRF + UMLS [34]	0.839	0.688	0.756
DNorm [3]	0.822	0.775	0.798
C-Bi-LSTM-CRF [34]	0.848	0.761	0.802
TaggerOne [22]	0.835	0.796	0.815
TaggerOne(+ normalization) [22]	0.851	0.808	0.829
DNER [24]	0.853	0.833	0.843
SBLC	0.866	0.858	0.862

ISSN: 1472-6947