word alignment 0.002123288
ibm model 0.002079458
alignment data 0.001990228
alignment models 0.001963108
markov model 0.0018648
translation models 0.00184129
convex model 0.001835161
english word 0.00175508
translation alignment 0.001738818
data sentence 0.001652985
data ibm 0.0016311580000000002
english data 0.00162202
data set 0.001614763
training data 0.001604725
ibm models 0.001604038
french word 0.001590413
word level 0.001564907
model 0.00150821
null word 0.001502397
bilingual word 0.001492272
english words 0.00147642
alignment variable 0.001464047
test data 0.001439681
second data 0.0014393610000000001
several alignment 0.0014280270000000001
standard alignment 0.001426652
ibm translation 0.001379748
hansards data 0.001362134
convex models 0.001359741
development data 0.001337784
previous alignment 0.001320893
french words 0.001311753
trial data 0.001308325
hansard data 0.001308325
target words 0.001293521
supervised models 0.001292994
objective function 0.0012690969999999998
translation parameters 0.001267643
complex alignment 0.001248283
alignment datasets 0.001247829
alignment workshop 0.001238339
likely alignment 0.001235838
final algorithm 0.0012018559999999998
new problem 0.001190513
optimization algorithm 0.001188467
translation pair 0.001187642
labeled alignment 0.00117834
english sentence 0.001155185
optimization method 0.001155019
first dataset 0.001146437
possible alignments 0.0011303420000000001
gradient method 0.001126865
training ibm 0.001116063
full translation 0.001113645
subgradient method 0.001110398
translation systems 0.001109356
machine translation 0.001103985
training set 0.001099668
translation mod 0.00109179
modern translation 0.001074338
optimization problem 0.001068001
chine translation 0.0010672049999999999
additional constraints 0.0010663040000000001
problem figure 0.001066236
translation quality 0.001065176
timization algorithm 0.0010548649999999999
algorithm opti 0.001051059
simple form 0.001045133
new variable 0.0010434670000000002
jective function 0.001038831
models 0.00103279
concave function 0.0010248269999999999
dataset results 0.001021575
same level 0.00101656
many training 0.001010877
first iteration 0.001009304
many results 0.001003374
parameter values 9.99141E-4
tion problem 9.97357E-4
local maximum 9.91884E-4
similar performance 9.910420000000001E-4
french sentence 9.90518E-4
first convex 9.89125E-4
original ibm 9.86393E-4
ing ibm 9.80323E-4
original training 9.5996E-4
ing training 9.5389E-4
standard relaxation 9.50705E-4
simplex constraints 9.44674E-4
test set 9.34624E-4
mization problem 9.32255E-4
ibm paper 9.31759E-4
alignment 9.30318E-4
bilingual corpus 9.29245E-4
intractible problem 9.29114E-4
optimization methods 9.2567E-4
first train 9.22164E-4
sentence pairs 9.173219999999999E-4
words 9.1431E-4
first pass 9.120809999999999E-4
