Why is each kana mora or each kanji character considered a separate word when searching? Here is an image of the rsults of me searching for よろしく on meta. You can see that the first three results have the whole phrase, but after that, it picks up any question/answer that has any of the characters in it. If you look under the "Advanced Search Tips", it says you can use double-quotes to search for an exact word or phrase. This works fine for English text, but again fails for Japanese text.

This makes searching in Japanese worthless, especially on the main site. Is there something that can be done about this?

searching meta

  • I agree. At the moment, I think it's still best to use Google search: site:japanese.stackexchange.com "よろしく" – snailplane Jun 28 '13 at 11:58
  • We don't have an estimate on how much effort it'd take to change/fix this at the moment. We are currently working on better supporting non-English alphabets, though, so we'll likely revisit this while we're at it. – Adam Lear Sep 26 '13 at 20:47

This can be fixed by making the searching algorithm refuse to split any CJK characters unless a space is present; even enclosing the characters in quotes, the searching expects a space after the quoted characters (due to it being focused on Latin sentence structure), and so gives incorrect results.

You could probably this and a dev on Stack Exchange will see it.

Small oversight due to language barriers, I think.

  • 1
    Do you mean to simply add the feature-request to this question, or open a new topic somewhere else? – istrasci Jul 24 '13 at 15:35
  • @istrasci I'll ping a dev about it, but I'd add it to the question – Amelia Jul 24 '13 at 16:22
  • Actually fixing search would require some language awareness, I think. You can't insist on spaces, nor can you always match individual characters. (Maybe when the Japanese Stack Overflow is launched they'll license some Japanese-aware search technology.) – snailplane May 5 '14 at 3:24

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .