Content Order-Controllable MR-to-Text
Content order is critical in natural language generation (NLG) for emphasizing the focus of a generated text passage. In this paper, we propose a novel MR (meaning representation)-to-text method that controls the order of the MR values in a generated text passage based on the given order constraints...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10320352/ |
| _version_ | 1827650375763623936 |
|---|---|
| author | Keisuke Toyama Katsuhito Sudoh Satoshi Nakamura |
| author_facet | Keisuke Toyama Katsuhito Sudoh Satoshi Nakamura |
| author_sort | Keisuke Toyama |
| collection | DOAJ |
| description | Content order is critical in natural language generation (NLG) for emphasizing the focus of a generated text passage. In this paper, we propose a novel MR (meaning representation)-to-text method that controls the order of the MR values in a generated text passage based on the given order constraints. We use an MR-text dataset with additional value order annotations to train our order-controllable MR-to-text model. We also use it to train a text-to-MR model to check whether the generated text passage correctly reflects the original MR. Furthermore, we augment the dataset with synthetic MR-text pairs to mitigate the discrepancy in the number of non-empty attributes between the training and test conditions and use it to train another order-controllable MR-to-text model. Our proposed methods demonstrate better NLG performance than the baseline methods without order constraints in automatic and subjective evaluations. In particular, the augmented dataset effectively reduces the number of deletion, insertion, and substitution errors in the generated text passages. |
| first_indexed | 2024-03-09T20:15:22Z |
| format | Article |
| id | doaj.art-4e4533d7cb17476caaae2b64c4565c3b |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-03-09T20:15:22Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-4e4533d7cb17476caaae2b64c4565c3b2023-11-24T00:00:48ZengIEEEIEEE Access2169-35362023-01-011112935312936510.1109/ACCESS.2023.333413910320352Content Order-Controllable MR-to-TextKeisuke Toyama0https://orcid.org/0000-0002-2079-8035Katsuhito Sudoh1https://orcid.org/0000-0002-2122-9846Satoshi Nakamura2https://orcid.org/0000-0001-6956-3803Nara Institute of Science and Technology, Nara, Ikoma, JapanNara Institute of Science and Technology, Nara, Ikoma, JapanNara Institute of Science and Technology, Nara, Ikoma, JapanContent order is critical in natural language generation (NLG) for emphasizing the focus of a generated text passage. In this paper, we propose a novel MR (meaning representation)-to-text method that controls the order of the MR values in a generated text passage based on the given order constraints. We use an MR-text dataset with additional value order annotations to train our order-controllable MR-to-text model. We also use it to train a text-to-MR model to check whether the generated text passage correctly reflects the original MR. Furthermore, we augment the dataset with synthetic MR-text pairs to mitigate the discrepancy in the number of non-empty attributes between the training and test conditions and use it to train another order-controllable MR-to-text model. Our proposed methods demonstrate better NLG performance than the baseline methods without order constraints in automatic and subjective evaluations. In particular, the augmented dataset effectively reduces the number of deletion, insertion, and substitution errors in the generated text passages.https://ieeexplore.ieee.org/document/10320352/Controllable text generationdata augmentationdata-to-textmeaning representationnatural language generation |
| spellingShingle | Keisuke Toyama Katsuhito Sudoh Satoshi Nakamura Content Order-Controllable MR-to-Text IEEE Access Controllable text generation data augmentation data-to-text meaning representation natural language generation |
| title | Content Order-Controllable MR-to-Text |
| title_full | Content Order-Controllable MR-to-Text |
| title_fullStr | Content Order-Controllable MR-to-Text |
| title_full_unstemmed | Content Order-Controllable MR-to-Text |
| title_short | Content Order-Controllable MR-to-Text |
| title_sort | content order controllable mr to text |
| topic | Controllable text generation data augmentation data-to-text meaning representation natural language generation |
| url | https://ieeexplore.ieee.org/document/10320352/ |
| work_keys_str_mv | AT keisuketoyama contentordercontrollablemrtotext AT katsuhitosudoh contentordercontrollablemrtotext AT satoshinakamura contentordercontrollablemrtotext |