understanding and generation modules
</term>
mediated
by
a
<term>
language neutral meaning representation
#428The CCLINC Korean-to-English translation system consists of two core modules, language understanding and generation modules mediated by a language neutral meaning representation called a semantic frame.
users
</term>
has been extensively studied
by
the
<term>
natural language generation community
#979The issue of system response to users has been extensively studied by the natural language generation community, though rarely in the context of dialog systems.
generation systems
</term>
can be overcome
by
employing
<term>
machine learning techniques
#1021We show how research in generation can be adapted to dialog systems, and how the high cost of hand-crafting knowledge-based generation systems can be overcome by employing machine learning techniques.
<term>
word string
</term>
has been obtained
by
using a different
<term>
LM
</term>
. Actually
#1109The 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.
the basis of
<term>
feedback
</term>
provided
by
<term>
human judges
</term>
. We reconceptualize
#1361In this paper, we present SPoT, a sentence planner, and a new methodology for automatically training SPoT on the basis of feedback provided by human judges.
for a
<term>
language L
</term>
are directed
by
a guide which uses the shared
<term>
derivation
#1714The non-deterministic parsing choices of the main parser for a language L are directed by a guide which uses the shared derivation forest output by a prior RCL parser for a suitable superset of L .
shared
<term>
derivation forest
</term>
output
by
a prior
<term>
RCL parser
</term>
for a suitable
#1724The non-deterministic parsing choices of the main parser for a language L are directed by a guide which uses the shared derivation forest output by a prior RCL parser for a suitable superset of L .
engine
</term>
can be improved dramatically
by
incorporating an approximation of the formal
#1885I show that the performance of a search engine can be improved dramatically by incorporating an approximation of the formal analysis that is compatible with the search engine's operational semantics.
for a
<term>
spoken dialogue system
</term>
by
eliciting
<term>
subjective human judgments
#2066In this paper We experimentally evaluate a trainable sentence planner for a spoken dialogue systemby eliciting subjective human judgments.
language system domains
</term>
. Motivated
by
the success of
<term>
ensemble methods
</term>
#2308Motivated by the success of ensemble methods in machine learning and other areas of natural language processing, we developed a multi-strategy and multi-source approach to question answering which is based on combining the results from different answering agents searching for answers in multiple corpora.
performance gains from the
<term>
data
</term>
by
using
<term>
class-dependent interpolation
#3073In 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 databy using class-dependent interpolation of N-grams.
</term>
for learning
<term>
morphology
</term>
by
identifying
<term>
hubs
</term>
in an
<term>
#3163We describe a simple unsupervised technique for learning morphologyby identifying hubs in an automaton.
a
<term>
corpus
</term>
automatically tagged
by
the first
<term>
learner
</term>
. The resulting
#3372Then, a Hidden Markov Model is trained on a corpus automatically tagged by the first learner.
translingual reach into other
<term>
languages
</term>
by
leveraging
<term>
human language technology
#3630It gives users the ability to spend their time finding more data relevant to their task, and gives them translingual reach into other languagesby leveraging human language technology.
</term>
of
<term>
data objects
</term>
created
by
the
<term>
system
</term>
during each
<term>
#3703The operation of the system will be explained in depth through browsing the repository of data objects created by the system during each question answering session.
recall
</term>
on both systems . Motivated
by
these arguments , we introduce a number
#4094Motivated by these arguments, we introduce a number of new performance enhancing techniques including part of speech tagging, new similarity measures and expanded stop lists.
</term>
after each
<term>
user utterance
</term>
.
By
holding multiple
<term>
candidates
</term>
#4194Since multiple candidates for the understanding result can be obtained for a user utterance due to the ambiguity of speech understanding, it is not appropriate to decide on a single understanding result after each user utterance. By holding multiple candidates for understanding results and resolving the ambiguity as the dialogue progresses, the discourse understanding accuracy can be improved.
<term>
models
</term>
are automatically derived
by
<term>
decision tree learning
</term>
using
#4360Moreover, the models are automatically derived by decision tree learning using real dialogue data collected by the system.
real
<term>
dialogue data
</term>
collected
by
the
<term>
system
</term>
. We obtained reasonable
#4369Moreover, the models are automatically derived by decision tree learning using real dialogue data collected by the system.
further improve the
<term>
stemmer
</term>
by
allowing it to adapt to a desired
<term>
#4500Monolingual, unannotated text can be used to further improve the stemmerby allowing it to adapt to a desired domain or genre.