N04-2006 |
errors affected performance in
|
article generation
|
. 4.5 Article Generation We trained
|
N04-2006 |
. These sets were used in the
|
article generation
|
experiments . TRAINGEN : This
|
N04-2006 |
83.6 % accuracy . 3 Approach The
|
article generation
|
task constitutes one component
|
C90-2046 |
sentence have an effect on the
|
article generation
|
in different languages . Head-Specifier
|
N04-2006 |
disambiguate the arti - cles .
|
Article generation
|
, however , clearly depends on
|
W00-0708 |
an important role in automated
|
article generation
|
. 1 Introduction Article choice
|
P06-1031 |
method is the best performing
|
article generation
|
method . " she is the good student
|
N04-2006 |
accordingly . 5 Future Work 5.1
|
Article Generation
|
We would like to improve the
|
P15-1084 |
techniques . However , recent work on
|
article generation
|
( Banerjee et al. , 2014 ) has
|
N04-2006 |
TRAINGEN , then performed the
|
article generation
|
task on all test sets . Table
|
N04-2006 |
applied the log-linear model on the
|
article generation
|
task , using features drawn from
|
N04-2006 |
weights . 4.2 Training Sets for
|
Article Generation
|
We created three additional training
|
N09-1059 |
such as MT candidate selection ,
|
article generation
|
, and countability detection
|
P15-1043 |
content models for scientific survey
|
article generation
|
by annotating sentences from
|
P15-1043 |
content models for scientific survey
|
article generation
|
containing 3,425 sentences from
|
P15-2023 |
listed in Table 5 incorporated
|
article generation
|
and demonstrated its positive
|
P15-1043 |
introduction sentences for survey
|
article generation
|
as opposed to previous work .
|
D12-1097 |
anaphora resolution for Japanese and
|
article generation
|
for English during translation
|
D13-1139 |
presicions come from postediting (
|
article generation
|
) . In table 4 , we show the
|
N04-2006 |
missing articles . On the one hand ,
|
article generation
|
performance degraded significantly
|