Skip to main content

Table 5 The performance of BioBERT large cased v1.1 on the Validation Set

From: Precision information extraction for rare disease epidemiology at scale

Evaluation level

Entity class

Precision

Recall

F1

Entity-level

Overall

0.824

0.851

0.837

EPI

0.905

0.953

0.929

STAT

0.688

0.559

0.617

LOC

0.765

0.761

0.763

DATE

0.857

0.889

0.873

SEX

0.949

0.987

0.968

ETHN

0.571

0.848

0.683

Token-level

Overall

0.921

0.865

0.892

EPI

0.962

0.962

0.962

STAT

0.906

0.729

0.808

LOC

0.936

0.801

0.863

DATE

0.962

0.989

0.975

SEX

0.963

0.987

0.975

ETHN

0.559

0.868

0.680