Skip to main content

Table 4 Performance Comparison with CRF-based Methods on the Training Set with 10-fold Cross Validation

From: Family member information extraction via neural sequence labeling models with different tag schemes

Configuration

Precision

Recall

F1-score

Baseline

0.882

0.857

0.870a

CRF-Baseline

0.836

0.743

0.787

Side

0.902

0.855

0.878a

CRF-Side

0.865

0.753

0.805

Relation-side

0.883

0.854

0.869a

CRF-Relation-side

0.850

0.700

0.768

  1. a Indicates passing the significant test under the level of 0.001. The p-values for the three configurations are 0.000006, 0.00005, and 0.000000004 respectively