ACL RD-TEC 1.0 Summarization of N03-2016
Paper Title:
COGNATES CAN IMPROVE STATISTICAL TRANSLATION MODELS
COGNATES CAN IMPROVE STATISTICAL TRANSLATION MODELS
Authors: Grzegorz Kondrak and Daniel Marcu and Kevin Knight
Primarily assigned technology terms:
- algorithm
- approximation
- cognate identification
- computing
- czech-english translation
- decoder
- giza
- identification
- machine translation
- parameter tuning
- phonetic transcription
- sentence alignment
- statistical machine translation
- statistical training
- statistical translation
- thresholding
- training algorithm
- transcription
- transliteration
- tuning
- validation
- word alignment
Other assigned terms:
- alignment error rate
- approach
- baseline model
- bilingual dictionaries
- bilingual dictionary
- bilingual parallel corpora
- bitext
- bleu
- bleu metric
- bleu score
- bleu scores
- case
- characters
- coefficient
- corpora
- dice
- dictionaries
- dictionary
- english language
- english language model
- error rate
- experimental results
- function words
- ibm model
- language model
- language pair
- language pairs
- lcsr
- lexical items
- measure
- measures
- method
- multilingual corpus
- n-gram
- names
- orthographic similarity
- parallel corpora
- perplexity
- precision
- punctuation
- reference translation
- reference translations
- sentence
- sentences
- similarity measure
- similarity measures
- standard deviation
- statistical significance
- term
- test data
- test set
- toolkit
- training
- training corpus
- training data
- training set
- translation models
- translation quality
- translations
- word
- word alignments
- word pair
- word similarity
- words
