N12-1029 |
Correction With a competitive
|
comma restoration
|
system in place , we turn to
|
N03-1029 |
have a substantive advantage for
|
comma restoration
|
. In order to use more information
|
N03-1029 |
which we evaluate our alternative
|
comma restoration
|
models . Beeferman et al. present
|
N03-1029 |
provides significant improvement to
|
comma restoration
|
performance . As it turns out
|
N03-1029 |
at an opportunity for improved
|
comma restoration
|
. The ec distribution is especially
|
N12-1029 |
error correction as essentially a
|
comma restoration
|
task , we can we use largely
|
N12-1029 |
While the task remains similar to
|
comma restoration
|
, error correction in student
|
N03-1029 |
our extensions . The input to
|
comma restoration
|
is a sentence x = x1 ... xn of
|
N12-1029 |
results . 6 Annotation For the
|
comma restoration
|
task , we needed only to obtain
|
N12-1029 |
with the same tagger used in the
|
comma restoration
|
experiments . Because we approach
|
N03-1029 |
approach Treebank quality . As such ,
|
comma restoration
|
may stand as the first end-user
|
N03-1029 |
significant performance improvement in
|
comma restoration
|
. ( Figure 1 lists performance
|
N03-1029 |
Section 23 . Precision of the
|
comma restoration
|
was 71.1 % and recall 55.2 %
|
N03-1029 |
5 Related Work We compare our
|
comma restoration
|
methods to those of Beeferman
|
N12-1029 |
error correction as we did for
|
comma restoration
|
. We still employ CRFs and label
|
N12-1029 |
state-of-the-art performance on the task of
|
comma restoration
|
, beating previous systems '
|
N12-1029 |
whereas they were removed in the
|
comma restoration
|
task . For error cor - rection
|
N12-1029 |
both the error correction and
|
comma restoration
|
tasks . We also developed and
|
N12-1029 |
single pass , rather than just
|
comma restoration
|
, but do provide results based
|
N12-1029 |
which has proven to be useful in
|
comma restoration
|
tasks ( see e.g. Lu and Ng ,
|