Skip to main content
. 2014 Mar 4;9(3):e71511. doi: 10.1371/journal.pone.0071511

Table 4. Experimental results of the comparing systems for finding similar questions.

Models BC BCF upBC STM LDAC LDACF
AP 0.543 0.556 0.564 0.575 0.638 Inline graphic †
% AP improvements over
BC N/A +2.39 +3.87 +5.89 +17.50 +20.81
BCF N/A N/A +1.44 +3.42 +14.75 +17.99
upBC N/A N/A N/A +1.95 +13.12 +16.31
STM N/A N/A N/A N/A +10.96 +14.09
LDAC N/A N/A N/A N/A N/A +2.82
p@1 0.550 0.561 0.577 0.585 0.648 Inline graphic †
†

indicates the results of our proposed methods are statistical significance over the four baseline methods (within 0.95 confidence interval using the Inline graphic-test). The results of our proposed approach are in bold.