请求优化ai分割单词,不要分割日语中的外来语
Mark Kaufmann
I don't really understand the issue here. Katakana words are words just like any others. They are identified and can be learned. Can you please explain better what you mean or share some examples from Japanese?
牛奶燕麦不加糖
Mark Kaufmann日语小说里面,有大量的虚拟人名和地名,图中的セラス是一个完整的人名,但是ai分割词汇将它分为俩个部分,分割为セ ラス,以及下面的虚拟地名アライオン ,同样分为了俩部分,小说里的地名和角色名字,你知道的,是贯穿整部小说的,ai分割后几乎全是这样的,如果是未分割的时候,机器翻译会自动用中文翻译出一个人名,但是分割以后就不会,只会从分割出一个和文章完全不相关的一个词汇,外来语中这样的例子非常多
Ai 分割词汇非常有用,在平假名中,将单词和短语都可以整理出来,但是片假名不同,一个很长的句子中,可能只有一个片假名,而且最重要的是在日语中,片假名的单词从西方引入 (( (片假名在日语中完全是作为单独的词汇并且独立存在的))) 没有变形后多种多样的单词,只是单独的词汇, 所以说完全不需要ai分割,直接保持完整的单词就好了
Mark Kaufmann
牛奶燕麦不加糖Ahh..ok. That makes sense. The issue is poor word splitting. We hope to improve that over time. But, it isn't perfect so you may always have to edit sentences to fix issues from time to time. Or, you can edit the lesson for your imported lessons and use the Replace All feature to fix an issue everywhere if the same wrong split has been made throughout your text.
In general, this should improve over time as AI will be doing this for us more and more.