Kotonoha!

A simple, entirely client-side dictionary currently based on the JMdict dataset.

View on GitHub Pages! https://moshikoi.github.io/Kotonoha/

Features

• Dictionary based on JMdict and Wiktionary
• Sentence parser powered by UniDic
• Example sentences from massif.la
  • Sentences ordered by vocabulary knowledge, so sentences with fewer unknown words are ranked higher

Todo

• Better ordering of results
  • Multi-token consideration
    • For instance, currently you can only search for one token of おはよう|ござい|ます. This should be changed so you will instead search for any of おはよう|おはようござい|おはようございます, as well as their dictionary forms. The more tokens are matched, the higher the search score
  • Better token merging heuristics
• More granualar vocabulary knowledge
• Import/export of user data
• Unidic pitch accents
• Stablize the database generation tools

Deploying

Following this guide, to deploy the site to GitHub Pages, run the following

git subtree push --prefix pages origin gh-pages

Licence information

The licences for libraries used are copied into the licences folder, with file names corresponding with the library.

sql.js is used under the MIT licence

Parts of UniDic 2.1.2 in a compressed format are included in the unidic folder, and are used under the terms of the BSD Licence file in that folder.