Abstract
We present a Portuguese↔English hybrid deep MT system based on an analysis-transfer-synthesis architecture, with transfer being done at the level of deep syntax, a level that already includes a great deal of semantic information. The system received a few months of development, but its performance is already similar to that of baseline phrase-based MT, when evaluated using BLEU, and surpasses the baseline under human qualitative assessment.
Cite
CITATION STYLE
Silva, J., Rodrigues, J., Gomes, L., & Branco, A. (2015). Bootstrapping a hybrid deep MT system. In ACL-IJCNLP 2015 - 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing, Proceedings of the 4th Workshop on Hybrid Approaches to Translation, HyTra 2015 (pp. 1–5). Association for Computational Linguistics (ACL). https://doi.org/10.18653/v1/w15-4101
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
