The other day, we developed budou that naturally delimits character phrases using deep learning. Japanese can be split by sharing, but it is troublesome to make it break more naturally.
However, it is mikan.js that it can be realized only on the client side without machine learning.
How to use mikan.js
When the width is wide.
When narrowing the most.
Separate into clauses to feel properly.
I understand by seeing mikan.js, but it is a regular expression or solved using the relationship between kanji and hiragana. Although it is machine learning in the sense of versatility, it may be easier to refine mikan.js in a very limited area called Japanese line break problem.
mikan.js is JavaScript open source software (MIT License).
No comments:
Post a Comment