Name
Abstract | ||
|---|---|---|
We present an experimental Machine Translation prototype system that is able to translate between Span- ish and English, using very basic linguistic resour ces. In our approach, no structural transfer rules are used to deal with structural divergences between the two la n- guages: the target corpus is the basis both for lex ical selection and for structure construction. Our strat egy emphasises modularity and language independence and, thus, is translatable to languages with very little NLP development. Our system is currently being developed in the framework of Metis-II (Vandeghinste et al., 2006). The goal of the Metis project is to achieve corpus-base d translation on the basis of a monolingual target co rpus and a bilingual dictionary only. The bilingual dict ionary functions as a flat translation model that provides n translations for each source word. The most probabl e translation given the context is then selected by c onsult- ing the statistical models built off the TL corpus 1. Clearly, syntactic divergences between the source and target languages are among the major challenges that this minimalist translation strategy faces. Tr ansfer systems typically address structural translation di ver- gences via explicit bilingual mapping rules, either hand- written or example-based. In the Spanish-English pr oto- type, we are able to do without a rule-based struct ural transfer component by handling translation divergen ces in the TL generation component. By pushing the treatment of translation mismatches to the TL end component of the system, we make the treatment independent of the source language and co n- sequently much more general. This solution is in li ne with other Generation intensive systems such as (Ha - bash & Dorr, 2002). Like us, they are able to dispe nse with expensive sophisticated resources for the Sour ce Language, however, unlike us, they need rich Target Language resources, such as lexical semantics, cate - gorial variation and subcategorisation frames. |
Year | Venue | DocType |
|---|---|---|
2007 | TMI | Conference |
Citations | PageRank | References |
1 | 0.35 | 0 |
Authors | ||
2 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Maite Melero | 1 | 22 | 9.29 |
| Toni Badia | 2 | 50 | 15.50 |