|
<term>
Communicator
</term>
participants are
|
using
|
. In this presentation , we describe the
|
#251
To support engaging human users in robust, mixed-initiative speech dialogue interactions which reach beyond current capabilities in dialogue systems, the DARPA Communicator program [1] is funding the development of a distributed message-passing infrastructure for dialogue systems which all Communicator participants are using. |
|
elicited from duplicating the experiment
|
using
|
<term>
machine translation output
</term>
.
|
#677
We tested this to see if similar criteria could be elicited from duplicating the experiment using machine translation output. |
|
<term>
word string
</term>
has been obtained by
|
using
|
a different
<term>
LM
</term>
. Actually ,
|
#1110
The oracle knows the reference word string and selects the word string with the best performance (typically, word or semantic error rate) from a list of word strings, where each word string has been obtained by using a different LM. |
|
combiner
</term>
with
<term>
hard decisions
</term>
|
using
|
the
<term>
reference
</term>
. We provide experimental
|
#1127
Actually, the oracle acts like a dynamic combiner with hard decisionsusing the reference. |
|
mimics the behavior of the
<term>
oracle
</term>
|
using
|
a
<term>
neural network
</term>
or a
<term>
decision
|
#1163
We suggest a method that mimics the behavior of the oracleusing a neural network or a decision tree. |
|
surprisingly close to what can be achieved
|
using
|
conventional
<term>
word-trigram recognition
|
#2247
The method combines domain independent acoustic models with off-the-shelf classifiers to give utterance classification performance that is surprisingly close to what can be achieved using conventional word-trigram recognition requiring manual transcription. |
|
performance gains from the
<term>
data
</term>
by
|
using
|
<term>
class-dependent interpolation
</term>
|
#3073
In this paper, we show how training data can be supplemented with text from the web filtered to match the style and/or topic of the target recognition task, but also that it is possible to get bigger performance gains from the data by using class-dependent interpolation of N-grams. |
|
NIST score
</term>
demonstrated the effect of
|
using
|
an out-of-domain
<term>
bilingual corpus
</term>
|
#3139
The two evaluation measures of the BLEU score and the NIST score demonstrated the effect of using an out-of-domain bilingual corpus and the possibility of using the language model. |
|
bilingual corpus
</term>
and the possibility of
|
using
|
the
<term>
language model
</term>
. We describe
|
#3148
The two evaluation measures of the BLEU score and the NIST score demonstrated the effect of using an out-of-domain bilingual corpus and the possibility of using the language model. |
|
<term>
Named Entity ( NE ) tagging
</term>
|
using
|
<term>
concept-based seeds
</term>
and
<term>
|
#3296
A novel bootstrapping approach to Named Entity (NE) taggingusing concept-based seeds and successive learners is presented. |
|
<term>
source interval projections
</term>
|
using
|
an underlying
<term>
word alignment
</term>
|
#3455
During training, the blocks are learned from source interval projectionsusing an underlying word alignment. |
|
subcategorization frame ( SCF )
</term>
distributions
|
using
|
the
<term>
Information Bottleneck
</term>
and
|
#3916
We describe a new approach which involves clustering subcategorization frame (SCF) distributions using the Information Bottleneck and nearest neighbour methods. |
|
derived by
<term>
decision tree learning
</term>
|
using
|
real
<term>
dialogue data
</term>
collected
|
#4362
Moreover, the models are automatically derived by decision tree learningusing real dialogue data collected by the system. |
|
proprietary
<term>
Arabic stemmer
</term>
built
|
using
|
<term>
rules
</term>
,
<term>
affix lists
</term>
|
#4551
Our resource-frugal approach results in 87.5% agreement with a state of the art, proprietary Arabic stemmer built using rules, affix lists, and human annotated text, in addition to an unsupervised component. |
|
</term>
.
<term>
Task-based evaluation
</term>
|
using
|
<term>
Arabic information retrieval
</term>
|
#4571
Task-based evaluationusing Arabic information retrieval indicates an improvement of 22-38% in average precision over unstemmed text, and 96% of the performance of the proprietary stemmer above. |
|
</term>
based on the results . The evaluation
|
using
|
another 23 subjects showed that the proposed
|
#5713
The evaluation using another 23 subjects showed that the proposed method could effectively generate proper referring expressions. |
|
of a
<term>
term aggregation system
</term>
|
using
|
each author 's text as a coherent
<term>
|
#6129
This paper proposes a new methodology to improve the accuracy of a term aggregation systemusing each author's text as a coherent corpus. |
|
</term>
in
<term>
unannotated text
</term>
by
|
using
|
a fully automatic sequence of
<term>
preprocessing
|
#7083
Furthermore, we present a standalone system that resolves pronouns in unannotated text by using a fully automatic sequence of preprocessing modules that mimics the manual annotation process. |
|
patterns
</term>
in
<term>
translation data
</term>
|
using
|
<term>
part-of-speech tag sequences
</term>
|
#7639
We describe a method for identifying systematic patterns in translation datausing part-of-speech tag sequences. |
|
sense disambiguation performance
</term>
,
|
using
|
standard
<term>
WSD evaluation methodology
|
#7814
We present the first known empirical test of an increasingly common speculative claim, by evaluating a representative Chinese-to-English SMT model directly on word sense disambiguation performance, using standard WSD evaluation methodology and datasets from the Senseval-3 Chinese lexical sample task. |