Skip to Main Content
Table 6 

Results for P@100, P@500, and the AP for various ranking measures. The gold standard is extracted using the GENIA corpus. This corpus is also used for computing the measures.

MethodP@100P@500AP
upper baseline 1.000 1.000 1.0000 
lower baseline 0.713 0.713 0.7134 
frequency 0.790 0.750 0.7468 
  
t-test 0.790 0.750 0.7573 
C-value (pre-pruned) 0.880 0.846 0.8447 
NC-value (pre-pruned) 0.880 0.840 0.8405 
GM 0.590 0.662 0.6740 
MF (pre-pruned) 0.920 0.872 0.8761 
FGM (pre-pruned) 0.910 0.840 0.8545 
MF (post-pruned) 0.900 0.876 0.8866 
FGM (post-pruned) 0.900 0.900 0.8948 
  
DRUID 0.930 0.852 0.8663 
DRUID (using word2vec) 0.800 0.740 0.7352 
Uniqueness (using word2vec) 0.680 0.752 0.7283 
Incompleteness (using word2vec) 0.760 0.724 0.7375 
log(freq) ⋅ DRUID 0.970 0.860 0.8661 
MF(post-pruned) ⋅ DRUID 0.950 0.926 0.9241 
FGM(post-pruned) ⋅ DRUID 0.960 0.940 0.9262 
MethodP@100P@500AP
upper baseline 1.000 1.000 1.0000 
lower baseline 0.713 0.713 0.7134 
frequency 0.790 0.750 0.7468 
  
t-test 0.790 0.750 0.7573 
C-value (pre-pruned) 0.880 0.846 0.8447 
NC-value (pre-pruned) 0.880 0.840 0.8405 
GM 0.590 0.662 0.6740 
MF (pre-pruned) 0.920 0.872 0.8761 
FGM (pre-pruned) 0.910 0.840 0.8545 
MF (post-pruned) 0.900 0.876 0.8866 
FGM (post-pruned) 0.900 0.900 0.8948 
  
DRUID 0.930 0.852 0.8663 
DRUID (using word2vec) 0.800 0.740 0.7352 
Uniqueness (using word2vec) 0.680 0.752 0.7283 
Incompleteness (using word2vec) 0.760 0.724 0.7375 
log(freq) ⋅ DRUID 0.970 0.860 0.8661 
MF(post-pruned) ⋅ DRUID 0.950 0.926 0.9241 
FGM(post-pruned) ⋅ DRUID 0.960 0.940 0.9262 
Close Modal

or Create an Account

Close Modal
Close Modal