ACL RD-TEC 1.0 Summarization of P02-1038
Paper Title:
DISCRIMINATIVE TRAINING AND MAXIMUM ENTROPY MODELS FOR STATISTICAL MACHINE TRANSLATION
DISCRIMINATIVE TRAINING AND MAXIMUM ENTROPY MODELS FOR STATISTICAL MACHINE TRANSLATION
Authors: Franz Josef Och and Hermann Ney
Primarily assigned technology terms:
- algorithm
- alignment template model
- approximation
- automatic speech recognition
- classification
- computing
- dialogue act classification
- direct optimization
- direct translation
- discriminative training
- dynamic programming
- dynamic programming search
- entropy training
- generalized iterative scaling
- iterative scaling
- language modeling
- language translation
- language understanding
- likelihood approach
- machine translation
- machine translation system
- machine translation systems
- maximum approximation
- maximum entropy
- maximum entropy approach
- maximum entropy model
- maximum likelihood
- maximum likelihood approach
- modeling
- mt system
- mt systems
- n-best rescoring
- natural language translation
- natural language understanding
- optimization
- parallel training
- pattern recognition
- pruning
- recognition
- rescoring
- sampling
- search
- search algorithm
- single reference translation
- speech recognition
- speech translation
- statistical alignment
- statistical machine translation
- statistical machine translation system
- statistical mt
- statistical translation
- translation system
- translation systems
Other assigned terms:
- acoustic model
- alignment model
- alignment models
- alignment template
- appointment scheduling
- approach
- bayes decision rule
- bleu
- bleu score
- bleu scores
- case
- decision rule
- dialogue act
- dictionary
- distribution
- edit distance
- entropy
- entropy models
- error rate
- experimental results
- fact
- feature
- french
- generation
- grammatical features
- hypothesis
- information theory
- interpretation
- knowledge
- language model
- language models
- language understanding task
- lexical features
- lexicon
- lexicon model
- likelihood
- mapping
- maximum entropy models
- measure
- measures
- method
- model parameter
- model parameters
- multi-reference word
- mutual information
- n-best list
- natural language
- natural languages
- optimization criterion
- optimization problem
- parallel training corpus
- parameter values
- posterior
- posterior probability
- precision
- probabilistic models
- probability
- probability distribution
- probability distributions
- reference translation
- reference translations
- search problem
- search space
- semantic
- semantic features
- sentence
- sentence pair
- sentences
- source language
- source language sentence
- source-channel model
- speech translation task
- statistical alignment models
- statistics
- symbol
- target language
- target language sentence
- target sentence
- target word
- test corpus
- theoretical framework
- theory
- training
- training corpus
- training criterion
- translation hypothesis
- translation model
- translation models
- translation probability
- translation quality
- translation task
- translations
- trigram
- trigram language model
- understanding
- verb
- verb groups
- word
- word error rate
- word order
- word trigram
- words
