C04-1045 |
during the EM training of the
|
statistical alignment
|
models . The evaluation is done
|
C04-1005 |
+ + toolkit is used to perform
|
statistical alignment
|
. Thus , for each sentence pair
|
C04-1045 |
POS information for improving
|
statistical alignment
|
quality of the HMM-based model
|
C04-1045 |
dependencies into the training of the
|
statistical alignment
|
models . Existing statistical
|
E06-1005 |
systems is used for the unsupervised
|
statistical alignment
|
training . Thus , the decision
|
C02-1009 |
alignment - candidate . Then the
|
statistical alignment
|
model is to find the Bayesian
|
E06-1005 |
of confusion net - works . 2.1
|
Statistical Alignment
|
The word alignment is performed
|
E06-1005 |
translation hypotheses with an enhanced
|
statistical alignment
|
algorithm that explicitly models
|
C04-1045 |
to improve the quality of the
|
statistical alignments
|
by taking into account the interdependencies
|
J03-1002 |
word alignment . So far , refined
|
statistical alignment
|
models have in general been rarely
|
E14-2013 |
bilingual dictionaries or by learning
|
statistical alignment
|
models out of bilingual corpora
|
D15-1051 |
2012 ) , in order to improve the
|
statistical alignment
|
models and make them more expressive
|
J03-1002 |
various design decisions of our
|
statistical alignment
|
system and evaluate these on
|
J03-1002 |
perform a symmetrization of directed
|
statistical alignment
|
models . As evaluation criterion
|
E03-1007 |
POS information for improving
|
statistical alignment
|
quality is described in ( Toutanova
|
J03-1002 |
Systematic Comparison of Various
|
Statistical Alignment
|
Models </title> Josef Hermann
|
C00-2163 |
Conclusion We have evaluated vm ` ious
|
statistical alignment
|
models by conlparing the Viterbi
|
E09-1020 |
one-size-fits-all " approach generally used in
|
Statistical alignment
|
and translation . Several interesting
|
E12-1010 |
) for a comparison of various
|
statistical alignment
|
models . In our case however
|
C04-1006 |
lexicon symmetrization methods for
|
statistical alignment
|
models that are trained using
|