ACL RD-TEC 1.0 Summarization of N06-1002
Paper Title:
DO WE NEED PHRASES? CHALLENGING THE CONVENTIONAL WISDOM IN STATISTICAL MACHINE TRANSLATION
DO WE NEED PHRASES? CHALLENGING THE CONVENTIONAL WISDOM IN STATISTICAL MACHINE TRANSLATION
Authors: Chris Quirk and Arul Menezes
Primarily assigned technology terms:
- algorithm
- alignment algorithm
- beam decoder
- beam search
- channel modeling
- computational linguistics
- controlled experimentation
- decoder
- decoding
- decomposition
- dependency parser
- giza
- heuristic combination
- human language
- human language technology
- kneser-ney smoothing
- language modeling
- language technology
- lexical weighting
- lexicalization
- likelihood estimation
- local reordering
- machine translation
- maximum likelihood
- maximum likelihood estimation
- modeling
- modeling translation
- mt systems
- n-gram language modeling
- non-monotone translation
- optimization
- parallel parsing
- parser
- parsing
- partitioning
- phrasal reordering
- phrasal translation
- phrase-based translation
- probability estimation
- search
- segmentation
- sequence modeling
- single reference translation
- smoothing
- smt system
- statistical machine translation
- syntax-based translation
- translation systems
- treelet translation
- weighting
- word alignment
- word alignment algorithm
- word translation
- word-for-word translation
Other assigned terms:
- aligned sentence
- alignment template
- anchor
- approach
- association for computational linguistics
- bag of words
- beam
- bias
- bigram
- bleu
- bleu metric
- case
- cohesion
- concept
- conditional probabilities
- contextual information
- contextual knowledge
- corpora
- data sparsity
- dependency parse
- dependency structure
- dependency tree
- dependency trees
- estimation
- exact match
- feature
- french
- generative models
- grammar
- heuristic
- hmm model
- ibm model
- implementation
- index
- knowledge
- language model
- language models
- language pairs
- lexical items
- likelihood
- linguistic
- linguistic information
- linguistics
- local context
- log-linear combination
- log-linear model
- long distance dependencies
- mappings
- method
- model size
- modeling problem
- mtu n-gram model
- n-best list
- n-gram
- n-gram language model
- n-gram model
- n-gram models
- non-compositionality
- non-terminal symbol
- parallel corpora
- parallel corpus
- parse
- parsing problem
- part-of-speech
- part-of-speech tags
- perplexity
- phrase
- phrase-based system
- prepositions
- probabilities
- probability
- probability distributions
- process
- reference translation
- reordering
- sentence
- sentence pair
- sentences
- sequence modeling problem
- software documentation
- source language
- source sentence
- sparse data
- statistical models
- substring
- symbol
- syntactic constituents
- syntactic information
- syntax
- tags
- target language
- target language model
- target sentence
- target word
- technology
- terms
- test set
- trained model
- training
- training data
- translation models
- translation pair
- translation pairs
- translation probabilities
- translation probability
- translation quality
- translation rules
- translations
- tree
- tree representation
- tree structure
- trees
- unigram
- unigram model
- word
- word alignment links
- word alignments
- word classes
- word count
- word order
- word sequence
- words
