Revisiting Multi-Domain Machine Translation
AbstractWhen building machine translation systems, one often needs to make the best out of heterogeneous sets of parallel data in training, and to robustly handle inputs from unexpected domains in testing. This multi-domain scenario has attracted a lot of recent work that fall under...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
The MIT Press
2021-01-01
|
| Series: | Transactions of the Association for Computational Linguistics |
| Online Access: | https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00351/97775/Revisiting-Multi-Domain-Machine-Translation |
| _version_ | 1818193481756049408 |
|---|---|
| author | MinhQuang Pham Josep Maria Crego François Yvon |
| author_facet | MinhQuang Pham Josep Maria Crego François Yvon |
| author_sort | MinhQuang Pham |
| collection | DOAJ |
| description |
AbstractWhen building machine translation systems, one often needs to make the best out of heterogeneous sets of parallel data in training, and to robustly handle inputs from unexpected domains in testing. This multi-domain scenario has attracted a lot of recent work that fall under the general umbrella of transfer learning. In this study, we revisit multi-domain machine translation, with the aim to formulate the motivations for developing such systems and the associated expectations with respect to performance. Our experiments with a large sample of multi-domain systems show that most of these expectations are hardly met and suggest that further work is needed to better analyze the current behaviour of multi-domain systems and to make them fully hold their promises. |
| first_indexed | 2024-12-12T00:47:05Z |
| format | Article |
| id | doaj.art-73a669e6e7db48679b3dc19578d47cc1 |
| institution | Directory Open Access Journal |
| issn | 2307-387X |
| language | English |
| last_indexed | 2024-12-12T00:47:05Z |
| publishDate | 2021-01-01 |
| publisher | The MIT Press |
| record_format | Article |
| series | Transactions of the Association for Computational Linguistics |
| spelling | doaj.art-73a669e6e7db48679b3dc19578d47cc12022-12-22T00:44:06ZengThe MIT PressTransactions of the Association for Computational Linguistics2307-387X2021-01-019173510.1162/tacl_a_00351Revisiting Multi-Domain Machine TranslationMinhQuang Pham0Josep Maria Crego1François Yvon2Université Paris-Saclay, CNRS, LIMSI, 91400, OrsaySYSTRAN, 5 rue Feydeau, 75002 Paris, France. josep.crego@systrangroup.comUniversité Paris-Saclay, CNRS, LIMSI, 91400, Orsay, France. francois.yvon@limsi.fr AbstractWhen building machine translation systems, one often needs to make the best out of heterogeneous sets of parallel data in training, and to robustly handle inputs from unexpected domains in testing. This multi-domain scenario has attracted a lot of recent work that fall under the general umbrella of transfer learning. In this study, we revisit multi-domain machine translation, with the aim to formulate the motivations for developing such systems and the associated expectations with respect to performance. Our experiments with a large sample of multi-domain systems show that most of these expectations are hardly met and suggest that further work is needed to better analyze the current behaviour of multi-domain systems and to make them fully hold their promises.https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00351/97775/Revisiting-Multi-Domain-Machine-Translation |
| spellingShingle | MinhQuang Pham Josep Maria Crego François Yvon Revisiting Multi-Domain Machine Translation Transactions of the Association for Computational Linguistics |
| title | Revisiting Multi-Domain Machine Translation |
| title_full | Revisiting Multi-Domain Machine Translation |
| title_fullStr | Revisiting Multi-Domain Machine Translation |
| title_full_unstemmed | Revisiting Multi-Domain Machine Translation |
| title_short | Revisiting Multi-Domain Machine Translation |
| title_sort | revisiting multi domain machine translation |
| url | https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00351/97775/Revisiting-Multi-Domain-Machine-Translation |
| work_keys_str_mv | AT minhquangpham revisitingmultidomainmachinetranslation AT josepmariacrego revisitingmultidomainmachinetranslation AT francoisyvon revisitingmultidomainmachinetranslation |