This is the latest version of this eprint.
Vu, D.N.L., Moosavi, N.S. orcid.org/0000-0002-8332-307X and Eger, S. (2022) Layer or representation space: what makes BERT-based evaluation metrics robust? In: Proceedings of the 29th International Conference on Computational Linguistics. 29th International Conference on Computational Linguistics (COLING 2022), 12-17 Oct 2022, Gyeongju, Republic of Korea. International Committee on Computational Linguistics , pp. 3401-3411.
Abstract
The evaluation of recent embedding-based evaluation metrics for text generation is primarily based on measuring their correlation with human evaluations on standard benchmarks. However, these benchmarks are mostly from similar domains to those used for pretraining word embeddings. This raises concerns about the (lack of) generalization of embedding-based metrics to new and noisy domains that contain a different vocabulary than the pretraining data. In this paper, we examine the robustness of BERTScore, one of the most popular embedding-based metrics for text generation. We show that (a) an embedding-based metric that has the highest correlation with human evaluations on a standard benchmark can have the lowest correlation if the amount of input noise or unknown tokens increases, (b) taking embeddings from the first layer of pretrained models improves the robustness of all metrics, and (c) the highest robustness is achieved when using character-level embeddings, instead of token-based embeddings, from the first layer of the pretrained model.
Metadata
Item Type: | Proceedings Paper |
---|---|
Authors/Creators: |
|
Copyright, Publisher and Additional Information: | © 2022 ACL. Licensed on a Creative Commons Attribution 4.0 International License. (https://creativecommons.org/licenses/by/4.0/) |
Dates: |
|
Institution: | The University of Sheffield |
Academic Units: | The University of Sheffield > Faculty of Engineering (Sheffield) > Department of Computer Science (Sheffield) |
Depositing User: | Symplectic Sheffield |
Date Deposited: | 07 Jun 2023 15:28 |
Last Modified: | 07 Jun 2023 17:35 |
Published Version: | https://aclanthology.org/2022.coling-1.300/ |
Status: | Published |
Publisher: | International Committee on Computational Linguistics |
Refereed: | Yes |
Related URLs: | |
Open Archives Initiative ID (OAI ID): | oai:eprints.whiterose.ac.uk:200109 |
Available Versions of this Item
-
Layer or representation space: what makes BERT-based evaluation metrics robust? (deposited 07 Jun 2023 15:21)
- Layer or representation space: what makes BERT-based evaluation metrics robust? (deposited 07 Jun 2023 15:28) [Currently Displayed]