Efficient Search for ITG

1
Efficient Search for ITG

• Hao Zhang and Daniel Gildea
• presented by Dakun Zhang
• 2007-02-05

2
Related Works

• Jason Eisner and Giorgio Satta. 1999. Eficient
  parsing for bilexical context-free grammars and
  head automaton grammars. In 37th Annual Meeting
  of the Association for Computational Linguistics.
• Richard Zens and Hermann Ney. 2003. A comparative
  study on reordering constraints in statistical
  machine translation. In Proceedings of the 40th
  Annual Meeting of the Association for
  Computational Linguistics, Sapporo, Japan.

3
Related Works

• Hao Zhang and Daniel Gildea. 2005. Stochastic
  lexicalized inversion transduction grammar for
  alignment. In Proceedings of the 43rd Annual
  Conference of the Association for Computational
  Linguistics (ACL-05), Ann Arbor, MI.
• Liang Huang, Hao Zhang, and Daniel Gildea. 2005.
  Machine translation as lexicalized parsing with
  hooks. In International Workshop on Parsing
  Technologies (IWPT05), Vancouver, BC.

4
ITG

• X ? e/f
• X ? e/e
• X ?e/f

• X ? YZ
• X ? ltYZgt

5
A Algorithm

• g(n) the cost of getting from the initial node
  to n.
• h(n) the estimate, according to the heuristic
  function, of the cost of getting from n to the
  goal node.
• f(n) g(n) h(n) intuitively, this is the
  estimate of the best solution that goes through
  n.

6
A Estimates for Alignment

• ITG constraints
• Xl,m,i,j
• IBM Model 1 estimate
• for the outside component

7
A Estimates for Alignment
8
A Decoding

• Alignment
• Find word level links between two fixed-order
  strings
• Decoding
• Input side is fixed
• Output side is a bag of words
• Find links and then reordered

9
A Decoding


O(n7)
? O(n6)
10
Experiments
11
Conclusion

• A search for alignments under ITG is efficient
• The hook trick can give further speedup for A
  decoding algorithm

12

• 3ks?
• Happy New Year!