ACL RD-TEC 1.0 Summarization of P95-1030
Paper Title:
NEW TECHNIQUES FOR CONTEXT MODELING
NEW TECHNIQUES FOR CONTEXT MODELING
Authors: Eric Sven Ristad and Robert G. Thomas
Primarily assigned technology terms:
- algorithm
- breadth-first search
- context modeling
- decoder
- encoding
- estimation algorithm
- greedy heuristic
- heuristic estimation
- heuristic model selection
- incremental model selection
- induction
- language modeling
- model estimation
- model selection
- model selection algorithm
- modeling
- parameter estimation
- parameter tying
- probability function
- search
- selection algorithm
- suffix tree
- transcription
- validation
Other assigned terms:
- alphabet
- approach
- axioms
- backoff
- brown corpus
- case
- characters
- community
- computational complexity
- conditional probabilities
- conditional probability
- context model
- context models
- corpus size
- data complexity
- dictionaries
- dictionary
- empirical results
- empty string
- entropy
- estimation
- events
- fact
- finite alphabet
- handwriting
- heuristic
- implementation
- knowledge
- language model
- language models
- large training
- leaf
- lemma
- maps
- mdl principle
- measures
- message
- method
- methodology
- minimum description length
- mixture models
- model complexity
- model parameters
- model performance
- model selection criterion
- n-gram
- n-gram model
- n-grams
- natural language
- natural language text
- ngram
- predictive power
- probabilities
- probability
- probability estimates
- probability model
- recursion
- selection rule
- substring
- suffix
- suffixes
- symbol
- symbols
- technique
- term
- terms
- test set
- testing corpus
- text
- text corpus
- theorem
- training
- training corpus
- training data
- transition probabilities
- tree
- trees
- trigram
- uniform probability
- unigram
- word
- word model
- word sequences
- words
