| Publication type: | Conference paper |
| Type of review: | Not specified |
| Title: | On the effectiveness of automated metrics for text generation systems |
| Authors: | von Däniken, Pius Deriu, Jan Milan Tuggener, Don Cieliebak, Mark |
| et. al: | No |
| Proceedings: | Findings of the Association for Computational Linguistics: EMNLP 2022 |
| Page(s): | 1503 |
| Pages to: | 1522 |
| Conference details: | The 2022 Conference on Empirical Methods in Natural Language Processing, Abu Dhabi, United Arab Emirates, 7-11 December 2022 |
| Issue Date: | 2022 |
| Publisher / Ed. Institution: | Association for Computational Linguistics |
| Language: | English |
| Subjects: | Text Generation; Artificial Intelligence (AI) |
| Subject (DDC): | 410.285: Computational linguistics |
| Abstract: | A major challenge in the field of Text Generation is evaluation, because we lack a sound theory that can be leveraged to extract guidelines for evaluation campaigns. In this work, we propose a first step towards such a theory that incorporates different sources of uncertainty, such as imperfect automated metrics and insufficiently sized test sets. The theory has practical applications, such as determining the number of samples needed to reliably distinguish the performance of a set of Text Generation systems in a given setting. We showcase the application of the theory on the WMT 21 and Spot-The-Bot evaluation data and outline how it can be leveraged to improve the evaluation protocol regarding the reliability, robustness, and significance of the evaluation outcome. |
| URI: | https://aclanthology.org/2022.findings-emnlp.108/ https://digitalcollection.zhaw.ch/handle/11475/27042 |
| Fulltext version: | Published version |
| License (according to publishing contract): | Licence according to publishing contract |
| Departement: | School of Engineering |
| Organisational Unit: | Centre for Artificial Intelligence (CAI) |
| Appears in collections: | Publikationen School of Engineering |
Files in This Item:
There are no files associated with this item.
Show full item record
von Däniken, P., Deriu, J. M., Tuggener, D., & Cieliebak, M. (2022). On the effectiveness of automated metrics for text generation systems [Conference paper]. Findings of the Association for Computational Linguistics: EMNLP 2022, 1503–1522. https://aclanthology.org/2022.findings-emnlp.108/
von Däniken, P. et al. (2022) ‘On the effectiveness of automated metrics for text generation systems’, in Findings of the Association for Computational Linguistics: EMNLP 2022. Association for Computational Linguistics, pp. 1503–1522. Available at: https://aclanthology.org/2022.findings-emnlp.108/.
P. von Däniken, J. M. Deriu, D. Tuggener, and M. Cieliebak, “On the effectiveness of automated metrics for text generation systems,” in Findings of the Association for Computational Linguistics: EMNLP 2022, 2022, pp. 1503–1522. [Online]. Available: https://aclanthology.org/2022.findings-emnlp.108/
VON DÄNIKEN, Pius, Jan Milan DERIU, Don TUGGENER und Mark CIELIEBAK, 2022. On the effectiveness of automated metrics for text generation systems. In: Findings of the Association for Computational Linguistics: EMNLP 2022 [online]. Conference paper. Association for Computational Linguistics. 2022. S. 1503–1522. Verfügbar unter: https://aclanthology.org/2022.findings-emnlp.108/
von Däniken, Pius, Jan Milan Deriu, Don Tuggener, and Mark Cieliebak. 2022. “On the Effectiveness of Automated Metrics for Text Generation Systems.” Conference paper. In Findings of the Association for Computational Linguistics: EMNLP 2022, 1503–22. Association for Computational Linguistics. https://aclanthology.org/2022.findings-emnlp.108/.
von Däniken, Pius, et al. “On the Effectiveness of Automated Metrics for Text Generation Systems.” Findings of the Association for Computational Linguistics: EMNLP 2022, Association for Computational Linguistics, 2022, pp. 1503–22, https://aclanthology.org/2022.findings-emnlp.108/.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.