Abstract
Recent work by Søgaard (2020) showed that, treebank size aside, overlap between training and test graphs (termed leakage) explains more of the observed variation in dependency parsing performance than other explanations. In this work we revisit this claim, testing it on more models and languages. We find that it only holds for zero-shot cross-lingual settings. We then propose a more fine-grained measure of such leakage which, unlike the original measure, not only explains but also correlates with observed performance variation. Code and data are available here: https://github.com/miriamwanner/reu-nlp-project- Anthology ID:
- 2022.findings-acl.230
- Volume:
- Findings of the Association for Computational Linguistics: ACL 2022
- Month:
- May
- Year:
- 2022
- Address:
- Dublin, Ireland
- Editors:
- Smaranda Muresan, Preslav Nakov, Aline Villavicencio
- Venue:
- Findings
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 2925–2934
- Language:
- URL:
- https://aclanthology.org/2022.findings-acl.230
- DOI:
- 10.18653/v1/2022.findings-acl.230
- Cite (ACL):
- Nathaniel Krasner, Miriam Wanner, and Antonios Anastasopoulos. 2022. Revisiting the Effects of Leakage on Dependency Parsing. In Findings of the Association for Computational Linguistics: ACL 2022, pages 2925–2934, Dublin, Ireland. Association for Computational Linguistics.
- Cite (Informal):
- Revisiting the Effects of Leakage on Dependency Parsing (Krasner et al., Findings 2022)
- PDF:
- https://preview.aclanthology.org/naacl-24-ws-corrections/2022.findings-acl.230.pdf
- Code
- miriamwanner/reu-nlp-project