GitHub - jungyeul/morphUD-korean (original) (raw)
Yet Another Format of Universal Dependencies for Korean (COLING 2020)
Yige Chen1*, Eunkyul Leah Jo2*, Yundong Yao2*, KyungTae Lim3, Miikka Silfverberg2, Francis M. Tyers4, and Jungyeul Park2 (2022). Yet Another Format of Universal Dependencies for Korean. Proceedings of the 29th International Conference on Computational Linguistics, 5432–5437. https://aclanthology.org/2022.coling-1.482 (*Yige Chen, Eunkyul Leah Jo, and Yundong Yao contributed equally.)
1The Chinese University of Hong Kong, Hong Kong, 2The University of British Columbia, Canada, 3Hanbat National University & TeddySum, South Korea, and 4Indiana University, USA.
Contents
- data: a dataset folder and their parsing results including morphUD, +morphUD and wordUD (original). morphUD and +morphUD datasets include their parsing results in the form of wordUD (for "fair" comparison).
- word2morph: a folder containing the scripts that converts the word-based UD format to the morpheme-based proposed format.
- morph2word: a folder containing the scripts that converts the morpheme-based proposed format back to the word-based UD format.
Results
| ko_gsd, wordUD | ko_gsd, morphUD | ko_kaist, wordUD | ko_kaist, morphUD | |
|---|---|---|---|---|
| UDPipe | 70.90 | 77.01 | 77.01 | 81.80 |
| Stanza | 84.63 (±0.18) | 84.98 (±0.20) | 86.67 (±0.17) | 88.46 (±0.14) |
References
@inproceedings{chen-EtAl:2022:COLING,
address = {Gyeongju, Republic of Korea},
author = {Chen, Yige and Jo, Eunkyul Leah and Yao, Yundong and Lim, KyungTae and Silfverberg, Miikka and Tyers, Francis M and Park, Jungyeul},
booktitle = {Proceedings of the 29th International Conference on Computational Linguistics},
month = {oct},
pages = {5432--5437},
publisher = {International Committee on Computational Linguistics},
title = {{Yet Another Format of Universal Dependencies for Korean}},
url = {https://aclanthology.org/2022.coling-1.482},
year = {2022}
}