SEE CHINESE — 500-CHARACTER LEARNING DATA

This release selects 500 unique traditional head characters, grouped in twenty
themed sets. Simplified forms are variants, not extra learning characters.

Sources downloaded on 12 October 2026:

1. CC-CEDICT, community-maintained Chinese-English dictionary, published by MDBG.
   Snapshot in its header: 2026-10-11T08:29:27Z.
   https://www.mdbg.net/chinese/dictionary?page=cedict
   https://cc-cedict.org/wiki/
   CEDICT copyright (C) 1997, 1998 Paul Andrew Denisowski.
   Licence: Creative Commons Attribution-ShareAlike 4.0 International.
   /content/licenses/CC-BY-SA-4.0.txt
   /data/word-source.txt contains the source entries used in this release.
   /data/catalog.js is adapted data under CC BY-SA 4.0: a selected reading,
   simplified spelling, short learner gloss, example and theme. See Chinese
   authored the focused lesson glosses, theme descriptions and short
   compositional phrases; the original dictionary entries are preserved.
   /content/selection.tsv, /content/themes.json and /content/overrides.json
   contain the editable lesson selection and reading choices, also CC BY-SA 4.0.
   /content/catalogue-config.json sets the target count and set size.

2. Make Me a Hanzi, by Shaunak Kishore and contributors.
   https://github.com/skishore/makemeahanzi
   Dictionary structure data is derived from Unihan and CJKlib.
   Licence: LGPL 3.0 or later, with the included Unicode notice.
   /content/licenses/LGPL-3.txt
   /content/licenses/GPL-3.txt
   /content/licenses/MMH-COPYING.txt
   /data/dictionary-source.ndjson preserves the selected source records.
   /data/structure.json is a modified subset: decomposition, stroke-to-component
   mappings and shorter component labels, compiled by See Chinese.

3. Modern stroke outlines and medians, Make Me a Hanzi graphics.txt, derived from
   Arphic PL KaitiM GB and Arphic PL UKai fonts. Copyright Arphic Technology.
   Licence: Arphic Public License, /content/licenses/ARPHIC.txt.
   /data/strokes/*.json contains the unmodified selected graphics records, split
   into twenty files for loading one set at a time. The paths are not historical
   glyph reconstructions. These data follow PRC stroke conventions; conventions
   can differ in other regions.

Rebuild this exact release with /scripts/build-content.mjs, the editable selection
files above, and the public data subsets. Full upstream files can be downloaded
using /scripts/fetch-sources.mjs when preparing a larger selection. These scripts
are provided under the MIT licence (/content/licenses/PIPELINE-MIT.txt).

Content source review completed on 14 October 2026:
All 500 selected character senses, headword readings, example meanings and
example readings were compared with the source records. The five translated
versions were reviewed against the selected senses. The five introductory
lessons, 26 phrases and 15 sentences were checked for meaning, phrasing,
context-specific glosses and character prerequisites. Eighteen catalogue
entries received reading, meaning or translation corrections.

Review method: AI-assisted source comparison and editorial correction by Codex.
Independent human sign-off is not recorded. The dated per-entry review record,
source links and checksums are available at /data/editorial-review.json.
Definitions use selected senses rather than every dictionary sense. The pinyin
shows dictionary citation tones; connected speech and regional readings can
differ. The roots-and-stems sense of 根莖 and the pointed-knife sense of 尖刀
also cite Taiwan Ministry of Education dictionaries:
https://dict.concised.moe.edu.tw/dictView.jsp?ID=16476&la=0&powerMode=0
https://pedia.cloud.edu.tw/Entry/Detail?title=%E5%B0%96%E5%88%80

The 21 illustrated story characters were checked against CUHK's Multi-function
Chinese Character Database. The borrowing row uses 來 and 萬. The traditional
老 / 考 comparison remains labelled as debated.
https://humanum.arts.cuhk.edu.hk/Lexis/lexi-mf/

Shape decomposition is not etymology. Component labels identify modern written
parts; a standalone meaning does not establish a part's role in every character.
The 21 characters linked to story lessons are a subset of the 500, not additional
characters. Original picture-to-character animations are explanatory illustrations.

The seven scenic backgrounds are AI-generated images created for See Chinese.
The learning app stores progress in the visitor's browser. Browser speech voices
are optional. No paid generation or external learning API is used at runtime.