Closed rabbit19981023 closed 1 year ago
No, MeCab can only use user dictionaries that are files, that's simply a design limitation.
For your use, you can automate the process of creating a file from a JSON dictionary. Alternately, you can avoid using a user dictionary and just do post-processing, which will work if the tokenization is already what you want and you just need to override kana.
Ok, I would like to do the post-processing.
Thanks a lot for your answer!
Hello,
I would like to write a simple tool for processing the japanese ebooks, and I want to wrap the service into a web ui.
One of the feature I would like to implement is allowing user to provide the custom kana tag for specific word, like 地名 or 人名.
I have read the source code, it seems like it uses mecab's api to build the user dict and specify it in
GenericTagger
, but I still wondering if just use a json object (it could be other format) is possible?