ACL RD-TEC 1.0 Summarization of W04-1115
Paper Title:
COMBINING PROSODIC AND TEXT FEATURES FOR SEGMENTATION OF MANDARIN BROADCAST NEWS
COMBINING PROSODIC AND TEXT FEATURES FOR SEGMENTATION OF MANDARIN BROADCAST NEWS
Primarily assigned technology terms:
- analysis pitch
- anaphora resolution
- asr transcription
- automatic speech recognition
- automatic topic segmentation
- boundary classification
- boundary detection
- boundary identification
- classification
- classifier
- classifier training
- classifiers
- combined classifier
- computing
- data analysis
- decision tree
- decision tree classifier
- detection and tracking
- discourse analysis
- encoding
- feature comparison
- identification
- information retrieval
- normalization
- preprocessing
- prosody-based identification
- recognition
- reporting
- sampling
- segmentation
- smoothing
- speaker change detection
- speaker identification
- speaker normalization
- speech recognition
- story segmentation
- subtopic segmentation
- summarization
- terminology
- text classification
- text-based segmentation
- topic detection
- topic detection and tracking
- topic segmentation
- transcription
- tree classifier
- tuning
- vector representation
- vector space model
- voicemail
- voting
- voting mechanism
- weighting
Other assigned terms:
- anaphora
- approach
- broadcast news
- broadcast news audio
- chinese words
- classification accuracy
- contextual features
- contour
- cosine similarity
- cue phrase
- cue phrase information
- data set
- discourse
- discourse markers
- discourse structure
- distribution
- document
- document frequency
- duration
- dutch
- error rate
- evaluations
- fact
- feature
- feature set
- feature sets
- feature types
- frame
- gesture
- gold standard
- information structure
- inverse document frequency
- labeling
- local context
- lookahead
- mandarin chinese
- measure
- measures
- n-gram
- news audio
- nist
- normalized word duration
- pause
- phoneme
- phoneme sequence
- phrase
- pitch
- pitch contour
- prosodic feature
- prosodic features
- prosodic information
- prosody
- representations
- segment boundaries
- segment boundary
- segments
- sentencelevel information
- signal
- silence
- similarity measure
- similarity measures
- size of the corpus
- speaker change
- speaking rate
- technique
- technology
- term
- term similarity
- test set
- text
- text similarity
- textual similarity
- tone
- topics
- training
- transcriptions
- tree
- unigram
- vector space
- word
- word duration
- word window
- words
