DMCA
cdec: A decoder, alignment, and learning framework for finite-state and context-free translation models (2010)
Cached
Download Links
| Venue: | In Proceedings of ACL System Demonstrations |
| Citations: | 134 - 53 self |
BibTeX
@INPROCEEDINGS{Dyer10cdec:a,
author = {Chris Dyer and Adam Lopez and Juri Ganitkevitch and Jonathan Weese and Hendra Setiawan and Ferhan Ture and Vladimir Eidelman and Phil Blunsom and Philip Resnik},
title = {cdec: A decoder, alignment, and learning framework for finite-state and context-free translation models},
booktitle = {In Proceedings of ACL System Demonstrations},
year = {2010}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
We present cdec, an open source framework for decoding, aligning with, and training a number of statistical machine translation models, including word-based models, phrase-based models, and models based on synchronous context-free grammars. Using a single unified internal representation for translation forests, the decoder strictly separates model-specific translation logic from general rescoring, pruning, and inference algorithms. From this unified representation, the decoder can extract not only the 1- or k-best translations, but also alignments to a reference, or the quantities necessary to drive discriminative training using gradient-based or gradient-free optimization techniques. Its efficient C++ implementation means that memory use and runtime performance are significantly better than comparable decoders. 1
Keyphrases
context-free translation model discriminative training efficient implementation present cdec statistical machine translation model gradient-free optimization technique phrase-based model memory use runtime performance single unified internal representation inference algorithm k-best translation synchronous context-free grammar general rescoring open source framework word-based model comparable decoder model-specific translation logic translation forest unified representation
