People cannot find other people because they do not know that a name is written with an umlaut or accent.
I think it is perfectly reasonable for a confluence user to expect that the search string 'rene' also finds the person 'René' and that a search for 'ozgur' also finds the person called 'Özgür'.

People suggesting that names with diacritic marks should be entered into Confluence without the diacritic marks, I kindly ask to crawl back in time a few decennia when the American Standards Association could use some help developing ASCII. The use of Unicode is not uncommon anymore, and this problem can be solved in a better way:

1. The preferred solution would be to implement a way to remove diacritic marks from a letter, so you can reduce a string with diacritic marks to a string without diacritic marks. A search string would then not only be compared to a persons name, but also to the reduced version of that name.

2. A work-around would be to create a new user property, besides 'fullname', that can contain alternative spellings of this persons name. Someone called René could enter 'Rene' as alternative spelling. A search strings would then not only be compared to a persons name, but also to the alternative spellings of that name.

More information:

• My support ticket describing the problem: https://support.atlassian.com/browse/CSP-40233
• Google demonstrating that you can strip the umlaut: http://www.google.nl/search?q=Dani%C3%ABl
• A solution in Perl: http://ahinea.com/en/tech/accented-translate.html