This is a Korean dictionary for the MeCab tokenizer, packaged for use in Python with mecab-python3 or fugashi.

We (at Luminoso) think this is probably a better way to find tokens in Korean words than just splitting the text on spaces would be. We also know that MeCab wasn't designed for Korean. But it gives us the kind of tokens we need better than other things we've tried.

One reason we're packaging this is to use it in wordfreq, which needs to have a reasonable set of consistent tokens so it can look up their frequencies, and already uses MeCab for Japanese.

If interoperability with things that already use MeCab is not your goal, you probably have better options for Korean NLP than this.

The dictionary data was created by Yongwoon Lee and Yungho Yu. We've included it as part of this package under the terms of the Apache License 2.0. The original dictionary can be found here.

The idea to package a MeCab dictionary as a Python repository, as well as the code structure for doing so, come from Paul McCann's ipadic package.

To install:

pip install mecab-ko-dic

To initialize with mecab-python3:

import MeCab
import mecab_ko_dic
tagger = MeCab.Tagger(mecab_ko_dic.MECAB_ARGS)
print(tagger.parse("안녕하세요세계."))

The data we use, and this code repository itself, are released under the Apache License 2.0. See the LICENSE.txt file included in this distribution.