Park J & Tyers F. A New Annotation Scheme for the Sejong Part-of-speech Tagged Corpus. In: Proceedings of the 13th Linguistic Annotation Workshop. Florence, Italy: Association for Computational Linguistics; 2019:195-202. https://www.aclweb.org/anthology/W19-4022.
sjmorph.model for udpipe (http://ufal.mff.cuni.cz/udpipe) is available at https://doi.org/10.5281/zenodo.3236528
We obtain 99.88% f1 score for segmentation and 94.75% accuracy for POS tagging for Sejong tag sets.
[October 2020] We update UPOS in the SJMorph model based on a new mapping table in kim-colineau:2020:LREC (to be announced).
[September 2020] We implement Korean NER based on the SJMorph model (to be announced).
[July 2020] Using sjmorph_v3.model, we obtain 99.88% f1 score for token segmentation and 94.77% accuracy for POS tagging for Sejong tag sets. We also fixed encoding problems in macOS. For the newest model of sjmorph.model, please contact jungyeul (dot) park (at) gmail (dot) com.