Beam Pruning Effects on Learning Stochastic Bracketing Inversion Transduction Grammar from Parallel Corpora

Markus Saers

11th International Conference on Parsing Technology (IWPT 2009)
Paris, France, 7th-9th October, 2009

---

Summary

We present a method for heavily pruning the estimation of Stochastic Bracketing Inversion Transduction Grammars. The estimated grammars are evaluated by building a standard Phrase-based statistical MT system on top of the alignments dictated by the Viterbi parse of the grammar. The trade-off between speed and translation quality is discussed, and compared to a standard word alignment system.

---

START Conference Manager (V2.56.8 - Rev. 780)