ACL RD-TEC 1.0 Summarization of W04-3009
Paper Title:
USING HIGHER-LEVEL LINGUISTIC KNOWLEDGE FOR SPEECH RECOGNITION ERROR CORRECTION IN A SPOKEN Q/A DIALOG
USING HIGHER-LEVEL LINGUISTIC KNOWLEDGE FOR SPEECH RECOGNITION ERROR CORRECTION IN A SPOKEN Q/A DIALOG
Authors: Minwoo Jeong and Byeongchang Kim and Gary Geunbae Lee
Primarily assigned technology terms:
- 6-fold cross validation
- algorithm
- asr error correction
- automatic speech recognition
- candidate generation
- channel error correction
- character recognition
- continuous speech recognition
- correction method
- cross validation
- database
- databases
- decoding
- dialog understanding
- dialogue system
- dialogue systems
- dictionary look-up
- dynamic programming
- dynamic programming search
- error correction
- error correction method
- error detection
- error detection and correction
- error reduction
- hidden markov
- hidden markov model
- information retrieval
- interfaces
- ir system
- language modeling
- lexical correction
- machine translation
- many-to-one mapping
- markov model
- matching
- maximum-likelihood
- maximum-likelihood estimation
- modeling
- morphological analysis
- natural language interface
- navigation
- ne recognition
- ne tagging
- nlp
- noisy channel model
- ocr error correction
- on-line education
- optical character recognition
- planning system
- post error correction
- qa system
- question answering
- reading
- recognition
- recognition error correction
- recognition system
- recognition systems
- recognition technology
- recognizer
- rule-based approach
- search
- search algorithm
- search process
- searching
- semantic correction
- semantic-oriented error correction
- sequence analysis
- smoothing
- speech error correction
- speech recognition
- speech recognition system
- speech recognition systems
- speech recognition technology
- speech recognizer
- speech transcription
- speech translation
- speech translation system
- spelling
- spelling correction
- spoken dialogue
- statistical language modeling
- statistical machine translation
- statistical method
- statistical mt
- tagging
- transcription
- translation system
- validation
- viterbi
- viterbi decoding
- viterbi search
- viterbi search algorithm
- word correction
Other assigned terms:
- acoustic model
- approach
- asr output
- car navigation
- co-occurrence
- co-occurrence information
- concept
- concepts
- conditional probability
- content words
- continuous speech
- dialog
- dialogue situation
- dictionaries
- dictionary
- discourse
- discourse structure
- distribution
- document
- document collection
- domain information
- domain knowledge
- edit distance
- error rate
- error reduction rate
- estimation
- events
- generation
- ibm model
- implementation
- keyboard input
- keyword
- knowledge
- knowledge base
- korean language
- korean speech
- language model
- language models
- lattice
- lexical entries
- lexical information
- lexical items
- lexical level
- lexico-semantic pattern
- lexicon
- likelihood
- linguistic
- linguistic knowledge
- mapping
- meaning
- measure
- method
- morphemes
- n-best list
- n-gram
- n-grams
- named entity
- names
- natural language
- noisy channel
- oneto-two channel probability
- ontology
- part-ofspeech
- penn treebank
- phoneme
- posterior
- posterior probability
- precision
- probabilistic information
- probabilities
- probability
- procedure
- process
- pronunciation
- pronunciation dictionary
- proper noun
- queries
- query
- query term
- question types
- recognition accuracy
- recognition errors
- recognition rate
- relation
- segments
- semantic
- semantic categories
- semantic category
- semantic class
- semantic classes
- semantic information
- semantic knowledge
- semantic representation
- semantic similarity
- semantic tags
- semantic type
- semantic types
- semantic-oriented approach
- sentence
- sentences
- speech input
- speech recognition error
- speech recognition errors
- spelling error
- structured language model
- syllable-based channel model
- syllables
- synonyms
- syntactic and semantic information
- syntactic level
- syntax
- tags
- technique
- technology
- term
- terms
- text
- toolkit
- training
- training corpus
- training data
- training set
- transcriptions
- transcripts
- transformation
- treebank
- twoto-one channel probability
- understanding
- user
- user query
- vocabulary
- word
- word error rate
- word lattice
- word model
- word sequence
- word sequences
- word-based model
- words
