learn.ifs.name

Statistical Phrase-Based Translation

Koehn et al., 2003, Statistical Phrase-Based Translation (Proceedings of the 2003 Human Language Technology Conference of the North American Chapter of ACL (HLT-NAACL 2003))

Subscriber deck — free during early access

2 cards in “4.1 Comparison of Core Methods” · whole deck

4.1 Comparison of Core Methods

  1. How did the three phrase extraction methods compare, and against what baseline?
    On German–English Europarl, from 10k to 320k sentence pairs: all consistent phrases (AP) slightly beat the joint model, syntactic phrases only were worse, and all phrase methods beat a word-based IBM Model 4 system. All systems improved with more data.
    • Koehn et al., 2003, Statistical Phrase-Based Translation (Proceedings of the 2003 Human Language Technology Conference of the North American Chapter of ACL (HLT-NAACL 2003)), §4.1, p. 51
  2. How does phrase table size grow, and how much do syntactic restrictions remove?
    Almost linearly with corpus size, because of the many singleton pairs (about 3.2M distinct pairs at 320k sentence pairs, maximum length 4). Restricting to syntactic phrases eliminated over 80% of pairs. Even the largest tables fit in a few hundred megabytes of RAM.
    • Koehn et al., 2003, Statistical Phrase-Based Translation (Proceedings of the 2003 Human Language Technology Conference of the North American Chapter of ACL (HLT-NAACL 2003)), §4.1, p. 51

Source

  • Philipp Koehn, Franz Josef Och and Daniel Marcu. 2003. Statistical Phrase-Based Translation. In Proceedings of the 2003 Human Language Technology Conference of the North American Chapter of the Association for Computational Linguistics (HLT-NAACL 2003), pp. 48–54. https://aclanthology.org/N03-1017/