ACL RD-TEC 1.0 Summarization of W04-2328
Paper Title:
MULTI-LEVEL DIALOGUE ACT TAGS
MULTI-LEVEL DIALOGUE ACT TAGS
Authors: Alexander Clark and Andrei Popescu-Belis
Primarily assigned technology terms:
- algorithm
- annotation program
- automatic annotation
- automatic tagging
- cl\/nlp
- classi cation
- classi er
- computational linguistics
- computing
- conversion procedure
- decoding
- dialogue act annotation
- dialogue management
- dialogue understanding
- feature selection
- forward-looking function
- hidden markov
- hidden markov model
- inference algorithm
- markov model
- maximum entropy
- meeting processing
- meeting processing and retrieval
- meeting recording
- multimodal dialogue
- nlp
- processing
- recognition
- recognizer
- retrieval systems
- search
- segmentation
- sequential inference
- speech recognition
- speech recognizer
- splitting
- tagging
- tagging process
- validation
Other assigned terms:
- acoustic information
- adjacency pair
- adjacency pair information
- annotation
- annotators
- asr accuracy
- case
- characters
- classi cation task
- community
- concepts
- contextual feature
- contextual features
- correlations
- damsl
- data sets
- dialog
- dialogue act
- dialogue acts
- dialogue management module
- dialogue structure
- dialogues
- dimensionality
- discourse
- distribution
- entropy
- fact
- feature
- feature set
- feature sets
- formalization
- gold standard
- grounding
- information sources
- inter-annotator agreement
- kappa
- knowledge
- language models
- lexical evidence
- lexical features
- likelihood
- linguistics
- linguists
- mapping
- mappings
- measure
- n-grams
- nlp application
- noise
- opinion
- particles
- probability
- probability distribution
- procedure
- process
- queries
- relation
- search space
- semantic
- semantic content
- sentences
- sources of information
- speech act
- speech acts
- switchboard corpus
- tags
- tagset
- term
- test set
- theories
- tokens
- topics
- training
- training set
- transcriptions
- transcripts
- understanding
- user
- utterance
- word
- words
