Statistical translation refers to an approach to machine translation that uses statistical models to estimate the most probable target-language translation of a source-language text based on patterns learned from bilingual or multilingual data [1][2]. In statistical machine translation (SMT), translation is commonly modeled by combining translation probabilities with a language model that evaluates the fluency or probability of candidate target-language sequences [1]. Model parameters are typically estimated from parallel corpora, in which source texts are aligned with their translations. Depending on the system architecture, statistical translation may operate at the word, phrase, or hierarchical syntactic level, using probabilistic methods to determine alignments and select among alternative translations [1][2][3].
- Statistical Machine Translation
- Probabilistic Models
- Parallel Corpora
- Phrase-Based Translation
References
- Brown, P.F.; Della Pietra, S.A.; Della Pietra, V.J.; Mercer, R.L. The Mathematics of Statistical Machine Translation: Parameter Estimation. Computational Linguistics 1993, 19, 263–311. https://aclanthology.org/J93-2003/
- Philipp Koehn; Franz Josef Och; Daniel Marcu. Statistical phrase-based translation; Association for Computational Linguistics (ACL): Kerrville, TX, United States, 2003; pp. 48-54. [CrossRef]
- Philipp Koehn; Kevin Knight. Feature-rich statistical translation of noun phrases; Association for Computational Linguistics (ACL): Kerrville, TX, United States, 2003; pp. 311-318. [CrossRef]