Skip to main content

Table 2 Local training/prediction vs. joint training and prediction over training/development sets; significant improvement made by the joint training models (IBT) on localization relationship (Loc) extraction

From: Structured learning for spatial information extraction from biomedical text: bacteria biotopes

  

LO

L+I

IBT-I

IBT+I

Bac.

P

0.959

0.959

0.991

0.972

 

R

0.993

0.994

0.970

0.978

 

F

0.976

0.976

0.980

0.975

Hab.

P

0.977

0.977

0.987

0.977

 

R

0.964

0.964

0.923

0.975

 

F

0.971

0.971

0.954

0.976

Loc.

P

0.188

0.20

0.311

0.318

 

R

0.274

0.268

0.584

0.580

 

F

0.223

0.229

0.406

0.411