8

If I remember correctly, on the main site the search is "accent-blind" in the sense that, for example, searching for José and Jose returns the same set of result. (Perhaps in different orders, exact matches might be considered more relevant.) Similarly for François and Francois.1

However, when I search in chat, searching for Jose does not return messages containing José. I am not sure how exactly are the messages returned when searching for José selected. But when I actually want to find messages containing the word José, I have to search for Jos.

Similarly, if I want to find messages containing the name François, I have to search for Fran or ois. (In short, this works as if the word was a part between two accented characters, or if there is only one then between the accented character and the beginning/end of the word.) I get completely different sets of results for François and for Francois.

If the word contains more accents, search seems to be more complicated. I have just tested in the sandbox that messages containing the name Désirée are found when searching for https://chat.stackexchange.com/search?q=sir&user=11287&room=1. They are not among the results if I search for Désirée or Desiree, that message is not among the search results.

  • Is it actually possible to search in chat for words including accented characters (such as the examples above)? Did I miss some trick I should have used?
  • If not, could some similar functionality as in the search of main be implemented.

I suppose that if there is name with accented characters, some users will use it in that way and some will transcribe it. (Especially if something is copy-pasted from somewhere, it is likely to contain accented characters.) If somebody is looking for past messages containing this name, AFAICT they have to do several searches.

1See, for example, this feature request which is marked : Search is too accent-sensitive.

You must log in to answer this question.

Browse other questions tagged .