ACL RD-TEC 1.0 Summarization of N04-1018
Paper Title:
DETECTING STRUCTURAL METADATA WITH DECISION TREES AND TRANSFORMATION-BASED LEARNING
DETECTING STRUCTURAL METADATA WITH DECISION TREES AND TRANSFORMATION-BASED LEARNING
Authors: Joungbum Kim and Sarah E Schwarm and Mari Ostendorf
Primarily assigned technology terms:
- algorithm
- automatic speech recognition
- automatic system
- boundary event detection
- classi cation
- clustering
- computing
- cross validation
- data collection
- decision tree
- decision tree training
- decision trees
- deterministic parser
- dis uency detection
- discourse marker
- editing
- entity detection
- error correction
- event detection
- extraction systems
- forward-backward algorithm
- hidden markov
- hidden markov model
- hmm-based approach
- information extraction
- information extraction systems
- ip detection
- language modeling
- language processing
- learner
- learning
- learning algorithm
- learning technique
- leave-one-out method
- linear interpolation
- machine learning
- markov model
- matching
- maximum entropy
- metadata detection
- modeling
- named entity detection
- natural language processing
- nlp
- nlp systems
- parser
- parsers
- parsing
- part-of-speech tagging
- pattern matching
- pos tagger
- post-processing
- predictor
- processing
- recognition
- recognizer
- repair
- rule learning
- score combination
- scoring
- scoring tool
- segmentation
- speech processing
- speech recognition
- speech recognizer
- speech-to-text
- spelling
- spelling correction
- sri language modeling
- su detection
- tagger
- tagging
- tagging-like metadata detection
- tbl training
- training process
- transcript evaluation
- transcription
- transformation-based learning
- tree-based modeling
- validation
- weighting
- word recognition
Other assigned terms:
- annotation
- approach
- bias
- case
- classi cation accuracy
- conversation
- conversational speech
- conversational telephone speech
- data consortium
- data sets
- decision tree model
- detection task
- discourse
- discourse markers
- distribution
- duration
- entropy
- error rate
- evaluation task
- evaluation test
- events
- experimental results
- fact
- feature
- information sources
- interpolation
- joint distribution
- knowledge
- language model
- language modeling toolkit
- language models
- leaf
- lexical features
- lexical information
- linguistic
- linguistic data
- linguistic data consortium
- machine prediction
- measure
- measures
- metadata
- method
- modeling decision
- modeling toolkit
- n-gram
- named entity
- natural language
- nist
- noise
- opinions
- part of speech
- part of speech tags
- part-of-speech
- pause
- pause duration
- pauses
- pos tag
- posterior
- posterior probability
- precision
- prediction accuracy
- probabilities
- probability
- process
- prosodic features
- prosodic information
- punctuation
- recognition errors
- segments
- semantic
- sentence
- sentence boundaries
- sentence level
- sentences
- silence
- slot
- sources of information
- speech data
- statistics
- structural information
- switchboard corpus
- system architecture
- system description
- tags
- technique
- term
- terms
- test data
- test set
- text
- token error rate
- tokens
- toolkit
- training
- training and test data
- training data
- transcript
- transcriptions
- transcripts
- tree
- tree model
- trees
- trigram
- utterance
- vocabulary
- vowel
- word
- word boundaries
- word boundary
- word error rate
- word error rates
- word fragments
- word sequence
- words
