ACL RD-TEC 1.0 Summarization of P95-1015
Paper Title:
COMBINING MULTIPLE KNOWLEDGE SOURCES FOR DISCOURSE SEGMENTATION
COMBINING MULTIPLE KNOWLEDGE SOURCES FOR DISCOURSE SEGMENTATION
Authors: Diane J. Litman and Rebecca J. Passonneau
Primarily assigned technology terms:
- algorithm
- boundary classification
- classification
- classification algorithm
- coding
- corresponding training
- cross-validation
- decision tree
- decision trees
- discourse segmentation
- feature representation
- hand tuning
- identification
- information retrieval
- learning
- learning algorithm
- learning algorithms
- learning method
- learning process
- learning program
- linking
- machine learning
- machine learning algorithm
- machine learning algorithms
- np algorithm
- predictor
- regression
- resampling
- segmentation
- spoken discourse
- tuning
Other assigned terms:
- anaphoric noun
- approach
- break
- case
- coherence
- cohesion
- contour
- corpora
- correlation
- cue phrase
- cue phrases
- cue words
- discourse
- discourse model
- discourse segment
- discourse structure
- discourse structures
- distribution
- duration
- error rate
- fact
- feature
- feature set
- generation
- human performance
- hypotheses
- hypothesis
- intention
- knowledge
- lexical cohesion
- lexical item
- linguistic
- linguistic features
- linguistic structure
- measures
- method
- methodology
- names
- noun phrase
- noun phrases
- paragraphs
- pause
- pause duration
- pauses
- phrase
- precision
- predictive power
- process
- pronoun
- pronouns
- prosodic features
- prosodic phrase
- prosody
- punctuation
- regression tree
- representative corpora
- rhetorical relations
- rhetorical structure
- rhetorical structure theory
- segment boundaries
- segment boundary
- segmental structure
- segments
- sentence
- signal
- technique
- term
- terms
- test corpus
- test set
- testing set
- text
- theory
- training
- training data
- training set
- transcripts
- tree
- trees
- understanding
- utterance
- verb
- verb arguments
- verb phrase
- word
- word features
- words
