Skip to main content
. 2019 Apr 9;19(Suppl 2):64. doi: 10.1186/s12911-019-0767-2

Table 5.

Test results of the two models with post-processing rules

Approach Strict index (%)
P R F1
BiLSTM-CRF 87.66 85.72 86.68
BiLSTM-CRF + rule 89.43 87.47 88.44
Attention-BiLSTM-CRF 87.75 86.77 87.26
Attention-BiLSTM-CRF + rule 89.61 88.58 89.09