Kana in question URLs should be transliterated

Question

The URLs generated for some of the questions are ridiculous (e.g. the URL of {～もあり、～もあり} vs {～もあって、～もあって}¹), and as such I think kana should be automatically transliterated for them so that there's some semblance of what the question is about. It would be nice if kanji could be handled as well, but it's understandable that it's much trickier to do so.

¹ https://japanese.stackexchange.com/questions/781/vs

Answer 1

We now encode titles using UTF-8 for the question and user slug. Let me know if you see anything strange as we are still testing this new feature.

Answer 2

Here is the usual way to encode UTF-8 into URLs from Stack Overflow.

I am pretty sure search engines are smart enough to decode and index this kind of URLs too (would need to test it though).

I would advise against any attempt at automatic transliteration. There are tools like Kakasi, but sometimes they get it wrong, and in those cases would confuse readers.

Answer 3

Strictly speaking, the part of the URL you are talking about is not an identifier (that is, it does not identify anything). It is completely ignored by the server. For example, the question 781 can be accessed by any of the following URLs:

• https://japanese.stackexchange.com/questions/781/vs
• https://japanese.stackexchange.com/questions/781
• https://japanese.stackexchange.com/questions/781/some-random-text
• https://japanese.stackexchange.com/questions/781/{～もあり、～もあり} vs {～もあって、～もあって}

This trick is intended to be part of search engine optimization. As you noticed, it is broken for non-Latin script. They can include Japanese text as it is (without transliteration) if they so wish. But I would not care too much about SEO tricks.

Added: I do not know (or care) whether this trick has any effect on search engines. The only thing that I know about the trick is that it is intended to be SEO.