MCE Training Techniques for Topic Identification of Spoken Audio Documents.
Timothy J. HazenPublished in: IEEE ACM Trans. Audio Speech Lang. Process. (2011)
Keyphrases
- spoken documents
- expert finding
- document set
- music scores
- document content
- minimum classification error
- document collections
- topic modeling
- information retrieval
- spoken document retrieval
- topic segmentation
- multi document summarization
- topic models
- training set
- web documents
- multimedia
- document retrieval
- broadcast news
- textual content
- information retrieval systems
- document clustering
- topic discovery
- xml documents
- query topic
- visual information
- relevant documents
- semantic information
- word frequency
- concept space
- news stories
- document representation
- metadata
- speech recognition
- training samples
- number of relevant documents
- feature selection
- language model
- keywords
- automatic summarization
- training corpus
- topic specific
- document level
- text corpora
- text classifiers
- document classification
- user interests