W03-0303 |
was shown to be effective in our
|
Chinese-English alignment
|
experiments . The first two passes
|
P09-1104 |
the best AER on the NIST 2002
|
Chinese-English alignment
|
data set . Furthermore , they
|
W03-0303 |
French-English and Romanian-English . For
|
Chinese-English alignment
|
, 365 sentence-pairs are randomly
|
P06-2112 |
distortion probability for head word in
|
Chinese-English alignment
|
. PrEJ ( Δj | Δj '
|
P09-1104 |
2005 ) . 6.2 Chinese NIST Results
|
Chinese-English alignment
|
is a much harder task than French-English
|
D10-1090 |
. Subse - quently , word level
|
Chinese-English alignments
|
are generated using the Berkeley
|
P11-1133 |
we reached a 77 % F-Measure for
|
Chinese-English alignment
|
using a classifier trained on
|
P09-1104 |
the lowest published error for
|
Chinese-English alignment
|
and an increase in downstream
|
D15-1210 |
Setup We evaluate our approach on
|
Chinese-English alignment
|
and translation tasks . The training
|
P13-1017 |
. We use the manually aligned
|
Chinese-English alignment
|
corpus ( Haghighi et al. , 2009
|
W03-0303 |
alignment tasks , and in addition to
|
Chinese-English alignment
|
. For Chinese - English and French-English
|
W02-1802 |
present some significant issues for
|
Chinese-English alignment
|
and term extraction for the construction
|
W05-1205 |
ITGs produces significantly lower
|
Chinese-English alignment
|
error rates than a syntactically
|
W03-0303 |
first describe our experiments on
|
Chinese-English alignment
|
, and then the results for the
|
P09-1105 |
MaxEnt aligner , we reduced the
|
Chinese-English alignment
|
error by 5 % and the Arabic-English
|
D11-1082 |
different starting aligners . 6.2
|
Chinese-English alignment
|
results Table ( 2 ) presents
|
N12-1078 |
examined 1/10 of all 5000 sampled
|
Chinese-English alignments
|
at random and found only 3 of
|
P09-1104 |
provide a much closer fit to true
|
Chinese-English alignments
|
than A1-1 . 3 Margin-Based Training
|