Puzikov, Yevgeniy ; Gardent, Claire ; Dagan, Ido ; Gurevych, Iryna (2019)
Revisiting the Binary Linearization Technique for Surface Realization.
The 12th International Conference on Natural Language Generation (INLG 2019). Tokyo, Japan (29.10.2019-01.11.2019)
doi: 10.18653/v1/W19-8635
Konferenzveröffentlichung, Bibliographie
Dies ist die neueste Version dieses Eintrags.
Kurzbeschreibung (Abstract)
End-to-end neural approaches have achieved state-of-the-art performance in many natural language processing (NLP) tasks. Yet, they often lack transparency of the underlying decision-making process, hindering error analysis and certain model improvements. In this work, we revisit the binary linearization approach to surface realization, which exhibits more interpretable behavior, but was falling short in terms of prediction accuracy. We show how enriching the training data to better capture word order constraints almost doubles the performance of the system. We further demonstrate that encoding both local and global prediction contexts yields another considerable performance boost. With the proposed modifications, the system which ranked low in the latest shared task on multilingual surface realization now achieves best results in five out of ten languages, while being on par with the state-of-the-art approaches in others.
| Typ des Eintrags: | Konferenzveröffentlichung |
|---|---|
| Erschienen: | 2019 |
| Autor(en): | Puzikov, Yevgeniy ; Gardent, Claire ; Dagan, Ido ; Gurevych, Iryna |
| Art des Eintrags: | Bibliographie |
| Titel: | Revisiting the Binary Linearization Technique for Surface Realization |
| Sprache: | Englisch |
| Publikationsjahr: | 5 September 2019 |
| Ort: | Tokyo, Japan |
| Verlag: | Association for Computational Linguistics |
| Buchtitel: | Proceedings of the 12th International Conference on Natural Language Generation |
| Veranstaltungstitel: | The 12th International Conference on Natural Language Generation (INLG 2019) |
| Veranstaltungsort: | Tokyo, Japan |
| Veranstaltungsdatum: | 29.10.2019-01.11.2019 |
| DOI: | 10.18653/v1/W19-8635 |
| URL / URN: | https://www.aclweb.org/anthology/W19-8635.pdf |
| Zugehörige Links: | |
| Kurzbeschreibung (Abstract): | End-to-end neural approaches have achieved state-of-the-art performance in many natural language processing (NLP) tasks. Yet, they often lack transparency of the underlying decision-making process, hindering error analysis and certain model improvements. In this work, we revisit the binary linearization approach to surface realization, which exhibits more interpretable behavior, but was falling short in terms of prediction accuracy. We show how enriching the training data to better capture word order constraints almost doubles the performance of the system. We further demonstrate that encoding both local and global prediction contexts yields another considerable performance boost. With the proposed modifications, the system which ranked low in the latest shared task on multilingual surface realization now achieves best results in five out of ten languages, while being on par with the state-of-the-art approaches in others. |
| Freie Schlagworte: | AIPHES,UKP_p_DIP,UKP_p_TGTOVE,FAZIT |
| Fachbereich(e)/-gebiet(e): | 20 Fachbereich Informatik 20 Fachbereich Informatik > Ubiquitäre Wissensverarbeitung DFG-Graduiertenkollegs DFG-Graduiertenkollegs > Graduiertenkolleg 1994 Adaptive Informationsaufbereitung aus heterogenen Quellen |
| Hinterlegungsdatum: | 18 Feb 2020 12:12 |
| Letzte Änderung: | 24 Mai 2024 13:00 |
| PPN: | |
| Export: | |
| Suche nach Titel in: | TUfind oder in Google |
Verfügbare Versionen dieses Eintrags
-
Revisiting the Binary Linearization Technique for Surface Realization. (deposited 11 Sep 2019 05:59)
- Revisiting the Binary Linearization Technique for Surface Realization. (deposited 18 Feb 2020 12:12) [Gegenwärtig angezeigt]
 |
Frage zum Eintrag |
Optionen (nur für Redakteure)
 |
Redaktionelle Details anzeigen |
Drucken
Impressum