Large-scale Exact Decoding: The IMS-TTT submission to WMT14

Abstract

We present the IMS-TTT submission to WMT14, an experimental statistical treeto-tree machine translation system based on the multi-bottom up tree transducer including rule extraction, tuning and decoding. Thanks to input parse forests and a "no pruning" strategy during decoding, the obtained translations are competitive. The drawbacks are a restricted coverage of 70% on test data, in part due to exact input parse tree matching, and a relatively high runtime. Advantages include easy redecoding with a different weight vector, since the full translation forests can be stored after the first decoding pass.

Findings of the 2014 Workshop on Statistical Machine Translation

“…University of Stuttgart / University of Munich (Quernheim and Cap, 2014) (Do et al, 2014) MANAWI-* Universität des Saarlandes (Tan and Pal, 2014) MATRAN Abu-MaTran Project: Prompsit / DCU / UA (Rubino et al, 2014) PROMT-RULE, PROMT-HYBRID PROMT RWTH RWTH Aachen STANFORD Stanford University (Neidert et al, 2014;Green et al, 2014) UA-* University of Alicante UEDIN-PHRASE, UEDIN-UNCNSTR University of Edinburgh (Durrani et al, 2014b) UEDIN-SYNTAX University of Edinburgh UU, UU-DOCENT Uppsala University (Hardmeier et al, 2014) Y-SDA Yandex School of Data Analysis (Borisov and Galinskaya, 2014) COMMERCIAL- [1,2] Two commercial machine translation systems ONLINE-[A,B,C,G] Four online statistical machine translation systems 4] Two rule-based statistical machine translation systems Table 2: Participants in the shared translation task. Not all teams participated in all language pairs.…”

Exact Decoding with Multi Bottom-Up Tree Transducers

“…Results are significantly worse compared to last year's system which used morphological enhancements such as compound splitting (Quernheim and Cap, 2014) and a phrase-based fallback system for sentences that the exact decoder could not handle. However, we should note that where the fallback system was not needed, we achieved a BLEU score of 16.7.…”

“…* This work was supported by Deutsche Forschungsgemeinschaft grant MA/4959/1-1. 1 The system presented in this paper is variant of the system presented at last year's workshop (Quernheim and Cap, 2014), without morphological enhancements.2 A translation is sensible if it is of linear size increase and can be computed by some (potentially copying) top-down tree transducer.…”