ACL RD-TEC 1.0 Summarization of P98-1047
Paper Title:
LEARNING A SYNTAGMATIC AND PARADIGMATIC STRUCTURE FROM LANGUAGE DATA WITH A BI-MULTIGRAM MODEL
LEARNING A SYNTAGMATIC AND PARADIGMATIC STRUCTURE FROM LANGUAGE DATA WITH A BI-MULTIGRAM MODEL
Authors: Sabine Deligne and Yoshinori Sagisaka
Primarily assigned technology terms:
- algorithm
- backoff smoothing
- backoff smoothing technique
- backoff technique
- class assignment
- clustering
- clustering algorithm
- database
- dialogue modeling
- dynamic programming
- estimation procedure
- forward-backward algorithm
- forward-backward training
- greedy algorithm
- grouping
- identification
- language modeling
- language understanding
- learning
- learning procedure
- linear interpolation
- maximum likelihood
- ml estimation
- modeling
- optimization
- parameter estimation
- parsing
- phrase clustering
- phrase retrieval
- pruning
- pruning strategy
- recognizer
- reestimation
- retrieving
- segmentation
- smoothing
- smoothing technique
- speech recognizer
- speech understanding
- stochastic language modeling
- structuring
- topic identification
- viterbi
- viterbi algorithm
- word bigram
- word grouping
- word prediction
Other assigned terms:
- ambiguity
- approach
- backoff
- bigram
- case
- class distribution
- cluster
- clustering procedure
- clusters
- conditional probability
- context free grammars
- convergence
- correlation
- correlations
- dialogues
- dictionaries
- distribution
- entropy
- estimation
- evaluation data
- formalism
- grammar
- grammar rules
- grammars
- heuristic
- interpolation
- knowledge
- language data
- language model
- language models
- likelihood
- local maximum
- meaning
- measure
- measures
- model size
- mutual information
- n-gram
- n-gram models
- n-grams
- ngram
- ngram model
- parse
- perplexity
- perplexity measure
- phrase
- phrase-based model
- pragmatic knowledge
- precision
- prediction accuracy
- priori
- probabilities
- probability
- procedure
- process
- recursion
- semantic
- sentence
- sentences
- speech act
- speech act tag
- stochastic model
- symbol
- technique
- term
- terms
- test data
- tokens
- toolkit
- training
- training and test data
- training corpus
- training data
- trigram
- trigram model
- understanding
- unigram
- unigram probability
- utterance
- vocabulary
- word
- word trigram
- words
