Quoted By:
Broadly, and making a lot of assumptions about the code base which I haven't looked at in detail, you would need to expand the type of value the tag fields can contain from a limited character set (lower chars only) to a wider one (perhaps full unicode). In parallell you would need to implement an expanded version of the search system that performs fuzzy searches for the expanded character set instead of the strict search which it currently performs (or just cast all the tags to lower case as you perform the search).
