Hypothesis 20 maria mary no dio not give did not no

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ed – p: score 1 to read o translation backtrack Miles Osborne Miles Osborne Machine Translation Machine Translation Machine Translation 16 February 2012 16 February 2012 16 February 2012 HypothesisExpansion Explosion! Hypothesis 20 Maria Mary no dio not give did not no did not give una bofetada a la bruja slap to by the witch green green witch a a slap slap to the to the slap e: witch f: -------*p: .182 e: f: --------p: 1 e: Mary f: *-------p: .534 verde the witch e: slap f: *-***---p: .043 e: did not f: **------p: .154 e: slap f: *****---p: .015 e: the f: *******-p: .004283 • Adding more hypothesis to find the best translation? §  Q: How much time e:green witch f: ********* p: .000271 §  Exponentially many translations, in length of source sentence ⇥ ExplosionNP-hard, just like for word translation models §  of search space §  So, we will use approximate search techniques! Miles Osborne Machine Translation 16 February 2012 Hypothesis Lattices Can recombine if: •  Last two English words match •  Foreign word coverage vectors match Decoder Pseudocode Initialization: Set beam Q={q0} where q0 is initial state with no words translated For i=0 … n-1 [where n in input sentence length] •  For each state q∈beam(Q) and phrase p∈ph(q) 1.  q’=next(q,p) [compute the new state] 2.  Add(Q,q’,q,p) [add the new state to the beam] Notes: •  ph(q): set of phrases that can be added to partial translation in state q •  next(q,p): updates the translation in q and records which words have been translated from input •  Add(Q,q’,q,p): updates beam, q’ is added to Q if it is in the top-n overall highest scoring partial translations Decoder Pseudocode Initialization: Set beam Q={q0} where q0 is initial state with no words translated For i=0 … n-1 [where n in input sentence length] •  For each state q∈beam(Q) and phrase p∈ph(q) 1.  q’=next(q,p) [compute the new state] 2.  Add(Q,q’,q,p) [add the new state to the beam] Possible State Representations: •  Full: q = (e, b, α), e.g. (“Joe did not give,” 11000000, 0.092) • ...
View Full Document

Ask a homework question - tutors are online