- Edited
Steps to reproduce:
Searching in Japanese with Japanese keywords
Expected behavior:
I would have expected a list of site in Japanese but many sites were in Chinese too which is irrelevant and unhelpful. Google does not have this issue.
Steps to reproduce:
Searching in Japanese with Japanese keywords
Expected behavior:
I would have expected a list of site in Japanese but many sites were in Chinese too which is irrelevant and unhelpful. Google does not have this issue.
I searched for 年金 自己投資
On my office PC using Chrome, I got responses from sites such as xueqiu.com (2nd result), m.wenxuecity.com (5th result), zh.ag.ny.gov (7th result), 1point3acres.com (9th result), etc. The 10th, 11th, 12th, 13th results are also Chinese.
I tried again on my personal PC using Edge and none of the results are Chinese.
I encountered the same problem when using kanji only query.
This shows that the search function in Japanese and Chinese is not running.
For example, I search "派遣" in kagi.com
https://kagi.com/search?q=%E6%B4%BE%E9%81%A3
I'm selecting Japan, but the search results contain a lot of Chinese results.
Google search result does not includes any chinese results If the search results area is Japan.
https://www.google.com/search?q=%E6%B4%BE%E9%81%A3
duckduckgo also does not include chinese in the search results if Japan is selected.
https://duckduckgo.com/?hps=1&q=%E6%B4%BE%E9%81%A3&atb=v398-3&ia=web
https://www.google.com/search?q=%E6%B4%BE%E9%81%A3
In Chinese and Japanese, the same kanji is used. it is difficult to judge by the word(kanji) itself.
Therefore, it is necessary to judge it by the content or lang etc...
https://github.com/greyblake/whatlang-rs
https://github.com/google/cld3
For example, https://zh.wikipedia.org/zh-hans/%E4%BA%BA%E5%8A%9B%E6%B4%BE%E9%81%A3 mark <html lang="zh-Hans">
and response header includes content-language: zh
.
At least this page can tell that it's not Japanese.
Vlad I think that the short-term solution will be to use another search engine like Google.
For example, "一" (number 1) search results in X includes chinese and japanese.
https://twitter.com/search?q=%E4%B8%80&f=live
However, X support lang:<lang>
and I add lang:ja
https://twitter.com/search?q=%E4%B8%80%20lang%3Aja&src=typed_query&f=live
If there is a way to specify the language of the search result in Kagi, I think I will use it.
I think that the language that users want is different for each user, so it is quite difficult to automatically determine.
For example, in my case, if the search term is English and Japanese, I want to prioritize English, but other users may not. However, I think that there are very few users who want both results to come out when Japanese and Chinese are mixed(This is because I have the experience that there are very few people who can understand both Japanese and Chinese).
Probably, priority-based ranking is necessary for each user rather than on or off if you try to automate it.
Hello Kagi Team and fellow users,
I have been a dedicated user of Kagi Search and have been thoroughly impressed with its precision and convenience. It's a service that I find worth paying for due to its outstanding quality.
However, I've encountered some issues when using the Orion browser, specifically with Japanese text. There are frequent cases where Japanese text is mistakenly recognized as Chinese, which makes it challenging to use the browser comfortably as a Japanese speaker. Because of this, I've been using Kagi as my default search engine in Brave instead.
I believe improving the Japanese language support in Orion would greatly benefit many users like myself who rely on Kagi's services. I sincerely hope this issue can be addressed in the near future.
Thank you for your time and attention.
SyotaNokemono591 Can you clarify what you mean?
Orion is browser, Kagi is search. This is forum for Kagi.
What exactly is the issue you have, how to reproduce it, and what do you want us to do?