# Third-party data notices

Hankan's application code is the authors'. The dictionary data it ships is not, and
carries share-alike obligations. The notices below are also surfaced in the app.

## Dictionary

Exactly one of these is used, depending on which source was vendored at build time.
`app/public/dict/manifest.json` records which one produced the shipped files.

### 한국어기초사전 (Basic Korean Dictionary) — preferred

Published by 국립국어원 (National Institute of Korean Language).

> 이 저작물은 크리에이티브 커먼즈 저작자표시-동일조건변경허락 2.0 대한민국
> 라이선스에 따라 이용할 수 있습니다.

한국어기초사전, 국립국어원 — CC BY-SA 2.0 KR
<https://creativecommons.org/licenses/by-sa/2.0/kr/>

Audio and image media referenced by the dictionary are **excluded** from that licence.
Hankan stores media URLs only and redistributes no media files.

### English Wiktionary, via kaikki.org — fallback

Extracted by wiktextract (Tatu Ylonen, LREC 2022; wiktextract itself is MIT).
Content is CC BY-SA 3.0 and GFDL, inherited from English Wiktionary.

<https://kaikki.org/dictionary/Korean/> · <https://en.wiktionary.org/>

## Frequency data

hermitdave/FrequencyWords, `content/2018/ko` — derived from OpenSubtitles2018 via OPUS.
Word list content is CC BY-SA 4.0; the project's code is MIT.

<https://github.com/hermitdave/FrequencyWords>

## A note on licence compatibility

CC BY-SA 2.0 KR and CC BY-SA 4.0 are not automatically compatible in a single derived
work. Hankan therefore keeps frequency data as a separately-attributed adjunct: it
populates the `lemma.freq_rank` column only, and is not merged into the dictionary
content itself. Both notices ship together.

Share-alike attaches to the derived dictionary data, not to Hankan's application code.