An Empirical Study on Automatic Post Editing for Neural Machine Translation
Automatic post editing (APE) researches aim to correct errors in the machine translation results. Recently, APE research has mainly been conducted in two directions: noise-based APE and adapter-based APE. This study poses three questions based on existing APE studies and conducted a verification. Th...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2021-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9528385/ |
| _version_ | 1819082802788827136 |
|---|---|
| author | Hyeonseok Moon Chanjun Park Sugyeong Eo Jaehyung Seo Heuiseok Lim |
| author_facet | Hyeonseok Moon Chanjun Park Sugyeong Eo Jaehyung Seo Heuiseok Lim |
| author_sort | Hyeonseok Moon |
| collection | DOAJ |
| description | Automatic post editing (APE) researches aim to correct errors in the machine translation results. Recently, APE research has mainly been conducted in two directions: noise-based APE and adapter-based APE. This study poses three questions based on existing APE studies and conducted a verification. The first is a question about the optimal APE research direction, and this has been figured out through a comparative analysis of the previous studies on noise-based APE and adapter-based APE. The second is about the substantial effectiveness of the bottleneck adapter layer (BAL) in adapter based APE. For the verification, various experiments on the different size of BAL has been conducted, and through these experiments, optimal approaches in adapter based APE has been proposed. For the last, this work raises a question about the reason why leveraging external knowledge is influential in APE. In this regard, we conducted several comparative experiments on the method of utilizing external data to APE training to achieve a better performance. The results revealed that the performance can be improved by applying the method of concatenating the external data with the existing data when training the APE model.Through deep analysis on these experiments, this work propose the optimal research direction in APE. |
| first_indexed | 2024-12-21T20:22:28Z |
| format | Article |
| id | doaj.art-e6f19f7523014d83bf9bd10b82a1ce08 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-21T20:22:28Z |
| publishDate | 2021-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-e6f19f7523014d83bf9bd10b82a1ce082022-12-21T18:51:28ZengIEEEIEEE Access2169-35362021-01-01912375412376310.1109/ACCESS.2021.31099039528385An Empirical Study on Automatic Post Editing for Neural Machine TranslationHyeonseok Moon0https://orcid.org/0000-0002-0841-4262Chanjun Park1https://orcid.org/0000-0002-7200-9632Sugyeong Eo2Jaehyung Seo3https://orcid.org/0000-0002-4761-9818Heuiseok Lim4https://orcid.org/0000-0002-9269-1157Department of Computer Science and Engineering, Korea University, Seoul, Republic of KoreaDepartment of Computer Science and Engineering, Korea University, Seoul, Republic of KoreaDepartment of Computer Science and Engineering, Korea University, Seoul, Republic of KoreaDepartment of Computer Science and Engineering, Korea University, Seoul, Republic of KoreaDepartment of Computer Science and Engineering, Korea University, Seoul, Republic of KoreaAutomatic post editing (APE) researches aim to correct errors in the machine translation results. Recently, APE research has mainly been conducted in two directions: noise-based APE and adapter-based APE. This study poses three questions based on existing APE studies and conducted a verification. The first is a question about the optimal APE research direction, and this has been figured out through a comparative analysis of the previous studies on noise-based APE and adapter-based APE. The second is about the substantial effectiveness of the bottleneck adapter layer (BAL) in adapter based APE. For the verification, various experiments on the different size of BAL has been conducted, and through these experiments, optimal approaches in adapter based APE has been proposed. For the last, this work raises a question about the reason why leveraging external knowledge is influential in APE. In this regard, we conducted several comparative experiments on the method of utilizing external data to APE training to achieve a better performance. The results revealed that the performance can be improved by applying the method of concatenating the external data with the existing data when training the APE model.Through deep analysis on these experiments, this work propose the optimal research direction in APE.https://ieeexplore.ieee.org/document/9528385/Automatic post editingneural machine translationadaptermultilingual pretrained language modelexternal knowledge |
| spellingShingle | Hyeonseok Moon Chanjun Park Sugyeong Eo Jaehyung Seo Heuiseok Lim An Empirical Study on Automatic Post Editing for Neural Machine Translation IEEE Access Automatic post editing neural machine translation adapter multilingual pretrained language model external knowledge |
| title | An Empirical Study on Automatic Post Editing for Neural Machine Translation |
| title_full | An Empirical Study on Automatic Post Editing for Neural Machine Translation |
| title_fullStr | An Empirical Study on Automatic Post Editing for Neural Machine Translation |
| title_full_unstemmed | An Empirical Study on Automatic Post Editing for Neural Machine Translation |
| title_short | An Empirical Study on Automatic Post Editing for Neural Machine Translation |
| title_sort | empirical study on automatic post editing for neural machine translation |
| topic | Automatic post editing neural machine translation adapter multilingual pretrained language model external knowledge |
| url | https://ieeexplore.ieee.org/document/9528385/ |
| work_keys_str_mv | AT hyeonseokmoon anempiricalstudyonautomaticposteditingforneuralmachinetranslation AT chanjunpark anempiricalstudyonautomaticposteditingforneuralmachinetranslation AT sugyeongeo anempiricalstudyonautomaticposteditingforneuralmachinetranslation AT jaehyungseo anempiricalstudyonautomaticposteditingforneuralmachinetranslation AT heuiseoklim anempiricalstudyonautomaticposteditingforneuralmachinetranslation AT hyeonseokmoon empiricalstudyonautomaticposteditingforneuralmachinetranslation AT chanjunpark empiricalstudyonautomaticposteditingforneuralmachinetranslation AT sugyeongeo empiricalstudyonautomaticposteditingforneuralmachinetranslation AT jaehyungseo empiricalstudyonautomaticposteditingforneuralmachinetranslation AT heuiseoklim empiricalstudyonautomaticposteditingforneuralmachinetranslation |