Could reading be made available in cases like 歪 | Bunpro where the sentences are already available? That might cover a few cases!
1 Like
I can’t find it on mobile
1 Like
Our app developer is working on it!
少々お待ちください 🙇♂️
1 Like
Hi !
I like the feature very much.
That said I’m wondering if this means you have given up on per-word clickability directly in the example sentences ?
That, I would regret.
2 Likes
We haven’t given up, it just takes very long to write an accurate parser, and even after parsing it will require human checks to make sure the content is correct.
Sorry for the delay with this, but it’s definitely still on the way!
4 Likes
Glad to hear it’s a work in progress ! It will be incredibly valuable once released.
May I ask which method you choose to build the parser, technically speaking. I’m just being a little curious. 
1 Like
I’m not sure exactly how Jake is doing it, but I think it uses Japanese tokenizers and a lot of custom logic.
No AI at the moment.
3 Likes
Can you not use an existing project instead of doing everything from scratch? I have seen several websites such as jisho that have the functionality to parse a Japanese sentence into its parts. Jisho uses mecab for parsing, and I suspect most other websites either make an api call to jisho or also use mecab directly. Both yomitan and mecab have parsers that are open source, and an api/wrappers that you could use. You could probably also just crowdsource the problem with a system where users can mark errors in the automatic parsing and if enough people mark the same change you could update it automatically.
Regardless of how it is done in the end, I am also looking forward to this feature.
2 Likes