The 2022 ReproGen shared task on reproducibility of evaluations in NLG: overview and results
Belz, AnyaORCID: 0000-0002-0552-8096, Shimorina, Anastasia, Popović, MajaORCID: 0000-0001-8234-8745 and Reiter, EhudORCID: 0000-0002-7548-9504
(2022)
The 2022 ReproGen shared task on reproducibility of evaluations in NLG: overview and results.
In: 15th International Conference on Natural Language Generation: Generation Challenges, 17-22 July 2022, Waterville, ME, USA.
Against a background of growing interest in reproducibility in NLP and ML, and as part of an ongoing research programme designed to develop theory and practice of reproducibility assessment in NLP, we organised the second shared task on reproducibility of evaluations in NLG, ReproGen 2022. This paper describes the shared task, summarises results from the reproduction studies submitted, and provides further comparative analysis of the results. Out of six initial team registrations, we received submissions from five teams. Meta-analysis of the five reproduction studies revealed varying degrees of reproducibility, and allowed further tentative conclusions about what types of eval- uation tend to have better reproducibility.
Proceedings of the 15th International Conference on Natural Language Generation: Generation Challenges.
.
Association for Computational Linguistics (ACL).
ReproHum project on Investigating Reproducibility of Human Evaluations in Natural Language Processing, funded by EPSRC (UK) under grant number EP/V05645X/1., ADAPT ´ SFI Centre for Digital Media Technology which is funded by Science Foundation Ireland under Grant 13/RC/2106., European Regional Development Fund (ERDF)
ID Code:
28369
Deposited On:
25 May 2023 14:50 by Maja Popovic. Last Modified 25 May 2023 14:50