training data 0.002010667
language model 0.0019885799999999998
english word 0.001912423
different language 0.001659142
learning model 0.001593117
compound word 0.001548214
translation model 0.001545601
compound words 0.001519394
long words 0.00151683
unknown words 0.001508658
complex words 0.001506137
supervised model 0.0014787429999999998
query data 0.0014574610000000002
fashionable words 0.001456736
decompounding model 0.001451424
noisy data 0.001450322
web data 0.0014496020000000001
finnish data 0.001382799
danish corpus 0.001309541
test language 0.001285187
different languages 0.001262713
additional information 0.001254991
language evaluation 0.0012517779999999998
many language 0.001237502
words 0.00122711
training sets 0.001218421
model 0.0012016
language models 0.0011820659999999998
cross language 0.001148451
result training 0.001147861
different lan 0.001145549
final training 0.00114214
information retrieval 0.001133676
mutual information 0.001127362
language processing 0.001113679
training instances 0.001105012
good results 0.0011039840000000001
different kinds 0.001101722
guage information 0.00109192
new features 0.001076896
previous methods 0.001069177
corpus 0.00104657
other languages 0.001043288
speech recognition 0.001035839
language morphemes 0.001032388
language cca 0.001027879
foreign language 0.0010224519999999999
same lan 0.001013578
common methods 0.001001937
method precision 9.90828E-4
other compound 9.450210000000001E-4
first step 9.44721E-4
splitting method 9.16566E-4
other hand 8.971000000000001E-4
additional features 8.87564E-4
training 8.74557E-4
detailed error 8.72496E-4
naive approach 8.48709E-4
information 8.44905E-4
possible way 8.29277E-4
correction system 8.24903E-4
possible number 8.2175E-4
large amount 8.1879E-4
parallel corpora 8.15721E-4
second step 8.13564E-4
full system 8.12641E-4
future work 8.04241E-4
supervised system 8.04023E-4
high recall 7.98139E-4
possible split 7.90751E-4
ing languages 7.88541E-4
language 7.8698E-4
possible compound 7.78481E-4
accuracy never 7.73518E-4
good recall 7.71618E-4
machine translation 7.6987E-4
previous list 7.653790000000001E-4
many texts 7.61567E-4
possible pair 7.60889E-4
german decompounding 7.531440000000001E-4
morphemes german 7.48728E-4
german decom 7.41387E-4
simple approaches 7.38561E-4
evaluation settings 7.34777E-4
speech 7.26477E-4
single decompounding 7.2126E-4
features most 7.157719999999999E-4
unigram features 7.10412E-4
good coverage 7.09952E-4
single decom 7.09503E-4
common source 7.07252E-4
several parts 7.004050000000001E-4
evaluation forum 6.96484E-4
several configurations 6.90582E-4
following way 6.86607E-4
standard sets 6.86131E-4
spelling correction 6.756500000000001E-4
kappa score 6.752990000000001E-4
human judges 6.69088E-4
previous instructions 6.63678E-4
