ACL RD-TEC 1.0 Summarization of N06-2021
Paper Title:
INITIAL STUDY ON AUTOMATIC IDENTIFICATION OF SPEAKER ROLE IN BROADCAST NEWS SPEECH
INITIAL STUDY ON AUTOMATIC IDENTIFICATION OF SPEAKER ROLE IN BROADCAST NEWS SPEECH
Primarily assigned technology terms:
- classification
- classifier
- classifiers
- clustering
- conditional likelihood
- conditional maximum entropy
- cross validation
- decision tree
- decoding
- dialog act classification
- entropy classifier
- feature extraction
- gaussian smoothing
- generative modeling
- generative modeling approach
- hidden markov
- hidden markov model
- hmm-based approach
- identification
- kneser-ney smoothing
- language processing
- linear interpolation
- markov model
- maxent
- maxent classifier
- maximum entropy
- maximum entropy classifier
- maximum entropy method
- maximum entropy model
- model training
- model training and testing
- modeling
- normalization
- processing
- question answering
- recognition
- role annotation
- role identification
- segmentation
- sequence decoding
- smoothing
- speaker clustering
- speaker role identification
- speaker segmentation
- speech recognition
- summarization
- tagging
- text normalization
- transcription
- validation
- video retrieval
- viterbi
- viterbi decoding
- weighting
- word segmentation
Other assigned terms:
- 10-fold cross validation
- acoustic information
- anchor
- anchors
- annotation
- approach
- background model
- benchmark
- broadcast news
- broadcast news corpus
- broadcast news data
- broadcast news speech
- classification accuracy
- classification performance
- classification task
- conditional probability
- context features
- contextual information
- corpora
- data set
- dependency relationship
- development set
- dialog
- dialog act
- duration
- entropy
- evaluations
- feature
- feature set
- grammar
- graphical representation
- hypothesis
- interannotator agreement
- interpolation
- key words
- knowledge
- labeling
- language model
- language processing applications
- lexical features
- likelihood
- maxent model
- method
- news corpus
- ngram
- nist
- opinions
- perplexity
- posterior
- probabilities
- probability
- prosodic features
- punctuation
- recognition model
- role labeling
- segments
- sentence
- sentences
- speaker change
- speaker identity
- structural information
- tags
- term
- test set
- text
- toolkit
- training
- training data
- training set
- transcriptions
- transcripts
- transition information
- transition probabilities
- tree
- utterance
- vocabulary
- word
- word sequence
- words
