ACL RD-TEC 1.0 Summarization of P04-1044
Paper Title:
COMBINING ACOUSTIC AND PRAGMATIC FEATURES TO PREDICT RECOGNITION PERFORMANCE IN SPOKEN DIALOGUE SYSTEMS
COMBINING ACOUSTIC AND PRAGMATIC FEATURES TO PREDICT RECOGNITION PERFORMANCE IN SPOKEN DIALOGUE SYSTEMS
Authors: Malte Gabsdil and Oliver Lemon
Primarily assigned technology terms:
- automatic classification
- classification
- classifiers
- confidence assessment
- cost assignment
- crossvalidation
- data collection
- dialogue evaluation
- dialogue management
- dialogue manager
- dialogue system
- dialogue systems
- feature weighting
- hypothesis reordering
- induction
- information pertaining
- language understanding
- learner
- learning
- learning system
- machine learner
- machine learning
- multi-modal dialogue
- multimodal dialogue
- natural language understanding
- nearest neighbors
- optimization
- optimization method
- parameter optimization
- parameter setting
- parser
- parsing
- predictor
- processing
- reading
- recognition
- recognizer
- route planning
- rule induction
- speech recognition
- speech recognition and synthesis
- speech recognizer
- splitting
- spoken dialogue
- spoken dialogue system
- spoken dialogue systems
- synthesis
- transcription
- tuning
- voting
- weighting
Other assigned terms:
- acoustic information
- anaphoric expressions
- approach
- buffer
- case
- clarification dialogue
- classification performance
- classification problem
- concept
- confidence rejection threshold
- confidence score
- confidence scores
- context-free grammar
- data structure
- determiners
- dialogue context
- dialogue move
- dialogues
- domain knowledge
- error rate
- evaluation metrics
- evaluation paradigm
- evaluations
- events
- f-score
- fact
- feature
- feature vectors
- frame
- generation
- gold standard
- grammar
- human operator
- hypotheses
- hypothesis
- information state
- information states
- interpretation
- knowledge
- labeling
- language model
- language models
- logical form
- logical forms
- measure
- method
- methodology
- modality
- multimodal command
- n-best list
- natural language
- noise
- oracle
- parameter settings
- parsing grammar
- procedure
- processing time
- pronoun
- pronouns
- recognition error rate
- recognition grammar
- relation
- reordering
- semantic
- semantic grammar
- signal
- speech input
- speech signal
- statistical significance
- statistics
- system performance
- technique
- terms
- test set
- theories
- training
- transcriptions
- tree
- tutorial dialogue
- understanding
- understanding component
- user
- user utterance
- user utterances
- utterance
- word
- word error rate
- words
