ACL RD-TEC 1.0 Summarization of N06-1036
Paper Title:
BACKOFF MODEL TRAINING USING PARTIALLY OBSERVED DATA: APPLICATION TO DIALOG ACT TAGGING
BACKOFF MODEL TRAINING USING PARTIALLY OBSERVED DATA: APPLICATION TO DIALOG ACT TAGGING
Authors: Gang Ji and Jeff Bilmes
Primarily assigned technology terms:
- algorithm
- automatic speech recognition
- backoff bigram
- backoff smoothing
- chain structuring
- classi cation
- clustering
- conditional maximum entropy
- da recognition
- data analysis
- decoding
- dialog translation
- discriminative modeling
- dynamic bayesian networks
- em algorithm
- em training
- error analysis
- expectation-maximization
- genetic algorithm
- hbm training
- hidden markov
- hidden markov model
- hidden markov models
- kneser-ney smoothing
- language modeling
- language processing
- linear regression
- markov model
- maximum entropy
- maximum-likelihood
- model training
- modeling
- natural language processing
- neural networks
- nlp
- normalization
- pattern recognition
- processing
- recognition
- regression
- search
- sequence analysis
- smoothing
- speech recognition
- state tying
- structuring
- tagging
- text generation
- training algorithm
- training method
- training procedure
- unsupervised clustering
- viterbi
- viterbi algorithm
- viterbi alignment
- viterbi training
- word prediction
Other assigned terms:
- annotation
- approach
- backoff
- backoff model
- baseline model
- bias
- bigram
- case
- chunk
- conditional probability
- convergence
- conversation
- corpora
- data set
- data sparseness
- development set
- dialog
- dialog act
- discourse
- discourse structures
- distribution
- entropy
- entropy models
- error rate
- estimation
- evaluations
- experimental results
- fact
- factored language model
- feature
- generation
- generative description
- generative model
- geometric distribution
- histogram
- hypothesis
- implementation
- joint distribution
- joint probability
- knowledge
- language model
- language model structure
- language models
- likelihood
- markov chain
- markov models
- maximum entropy models
- meaning
- method
- methodology
- model complexity
- model structure
- natural language
- noun phrase
- phrase
- probability
- probability distributions
- procedure
- prosody
- research topic
- semantic
- sentence
- sentences
- standard deviation
- statistics
- tagging accuracy
- tags
- technique
- test data
- test set
- text
- toolkit
- topics
- trained model
- training
- training corpus
- training data
- training set
- training time
- transition matrix
- transition probability
- trees
- understanding
- verb
- verb phrase
- word
- word distribution
- word model
- word sequence
- words
