... implement. The Backoff Erk model is
the best, using the Baseline for the majority of
decisions and backing off to the Erk smoothing
model when the Baseline cannot answer.
Figure 5 (shown on the next ... which of two verbs
was the correct predicate for a given noun object.
One verb v was the original from the source doc-
ument, and the other v
was randomly gener...
... and the former was used as the training
data and the latter as the development data. For
semi-CRFs, we used amis
3
for training the semi-
CRF with feature-forest. We used GENIA taggar
4
for POS-tagging ... than the system without it (the p-value is
less than 1.0 < 10
−4
). The result of the preceding
entity information improves the performance. On
the other han...
... vectors: Another
lesson of Tab. 3 is that the effect of the com-
position of the feature vectors can vary depend-
ing both on the task and on the size of the fea-
ture vector. The dramatic ... because some classes, such as L, are
very small. This problem was not so grave for the
LPE experiments because of the ceiling effect and
the small size of the c...
... the
contribution the features exemplified in one baseline
and six versions of the SVM model. The baseline is
defined only for the English part of the NP feature
set and measures the the contribution of the ... EXPERIENCER, THEME, BENEFICIARY.
Out of these instances, 74.81% use the preposition
of. In CLUVI, 11.71% of the examples were ver-
bal, from which the...
... lexicon of the tar-
get grammar,
4
and make use of the existing sets of
4
When the lexicon is less accurate, I can determine the
number of clusters using other algorithms (Hamerly, 2003).
SCFs for the ... vectors from the training
SCFs and the acquired SCFs for the words in the
testing SCFs. The number of the resulting data
objects was 8,679 for XTAG an...
... cases where the
query is composed of two or more sentences, we
compute the similarity between the document sen-
tence (s) and each of the query-sentences (q
i
) then
we take the average of the scores.
3 ... but
at the same time makes it not well suited for the se-
mantic trees (ST) defined in Section 3. For instance,
although the two STs of Figure 1 share most of...
... on the English part
of the bitexts and the Gigaword corpus of about
3.2 billion words. Therefore, it is likely that the
target language model includes at least some of
the translations of the ... amounts of parallel texts
to translate the source side of the non-
parallel corpus. The target side texts are
used, along with other corpora, in the lan-
guage model...
... is the number of sentences
in the sample, and the % column gives the sample
size as a percentage of the whole section.
We compare the CCG parser to the Berkeley
parser using the accurate mode of ... closer to the PTB than CCGbank, or due to their
conversion method. We leave the application of their method
to the CCG parser for future work.
to use the comple...
... layouts, there is a minority
strategy, used by 4% of the speakers (3 out of 72 cases
of the data of Linde (1974)) describing the layout in
the form of a map. The speaker first describes the
outside ... describe the layout of
their apartment. The vast majority of speakers used a
"tour strategy," which takes the hearer on an imaginary
tour of th...
... function of the CXXXC
motif of human Sco proteins could therefore be impli-
cated not only in the maturation of the Cu
A
site of
Cox2 but also in the maintenance of cellular copper
homeostasis.
The ... assembly of cbb
3
oxidase, but rather is
required for the maturation of the Cu
A
-containing
COX which is predominant for aerobic growth, thus
leaving open the...