The KAIST Korean Universal Dependency Treebank is generated by Chun et al., 2018 from the constituency trees in the KAIST Tree-Tagging Corpus.
This is a collaborative work by (in alphabetic order):
- Jinho Choi, Emory University
- Jayeol Chun, Emory University
- Na-Rae Han, University of Pittsburgh
- Jena D. Hwang, Institute for Human & Machine Cognition.
The project repository: https://github.com/emorynlp/ud-korean
- Building Universal Dependency Treebanks in Korean, Jayeol Chun, Na-Rae Han, Jena D. Hwang, and Jinho D. Choi. In Proceedings of the 11th International Conference on Language Resources and Evaluation, LREC'18, Miyazaki, Japan, 2018.
- 2022-11-15 v2.11
- Fixed right-headed apposition and non-projective punctuation.
- Symbols after numbers are units, not punctuation.
- Nouns cannot be attached as mark.
- Fixed: adverbially used nominals are obl, not advmod.
- Fixed: adverbially used verbs are advcl, not advmod.
- Positive copula is always lemmatized 이 so that the validator recognizes it.
- Fixed auxiliaries.
- Fixed: function words should be leaves.
- 2018-04-15 v2.2
- First release by Chun et al., 2018.
=== Machine-readable metadata (DO NOT REMOVE!) ================================ Data available since: UD v2.2 License: CC BY-SA 4.0 Includes text: yes Genre: news fiction academic Lemmas: converted from manual UPOS: converted from manual XPOS: converted from manual Features: not available Relations: converted from manual Contributors: Choi, Jinho; Han, Na-Rae; Hwang, Jena; Chun, Jayeol Contributing: here Contact: email@example.com =============================================================================== (Original treebank contributors: Choi, Key-Sun)