Skip to Main Content
Table 8 

Results of n-gram ranking on the medical data. Whereas the gold standard is extracted from the GENIA data set, the ranking measures as well as the frequency threshold for selecting the gold candidates are computed using the Medline corpus.

MethodP@100P@500AP
upper baseline 1.000 1.000 1.0000 
lower baseline 0.416 0.416 0.4161 
frequency 0.720 0.534 0.4331 
  
C-value (pre-pruned) 0.750 0.564 0.4519 
t-test 0.720 0.542 0.4483 
GM 0.210 0.272 0.3502 
MF (pre-pruned) 0.550 0.542 0.4578 
FGM (pre-pruned) 0.580 0.478 0.4200 
MF (post-pruned) 0.530 0.500 0.4676 
FGM (post-pruned) 0.490 0.446 0.4336 
  
DRUID 0.770 0.686 0.4608 
log(freq) ⋅ DRUID 0.860 0.720 0.4693 
GM ⋅ DRUID 0.770 0.634 0.4497 
MF(pre-pruned) ⋅ DRUID 0.730 0.634 0.4824 
MF(post-pruned) ⋅ DRUID 0.730 0.626 0.4889 
MethodP@100P@500AP
upper baseline 1.000 1.000 1.0000 
lower baseline 0.416 0.416 0.4161 
frequency 0.720 0.534 0.4331 
  
C-value (pre-pruned) 0.750 0.564 0.4519 
t-test 0.720 0.542 0.4483 
GM 0.210 0.272 0.3502 
MF (pre-pruned) 0.550 0.542 0.4578 
FGM (pre-pruned) 0.580 0.478 0.4200 
MF (post-pruned) 0.530 0.500 0.4676 
FGM (post-pruned) 0.490 0.446 0.4336 
  
DRUID 0.770 0.686 0.4608 
log(freq) ⋅ DRUID 0.860 0.720 0.4693 
GM ⋅ DRUID 0.770 0.634 0.4497 
MF(pre-pruned) ⋅ DRUID 0.730 0.634 0.4824 
MF(post-pruned) ⋅ DRUID 0.730 0.626 0.4889 
Close Modal

or Create an Account

Close Modal
Close Modal