Since a couple of months, CafeTran can give fuzzy matches for glossary entries. This is a very nice feature when e.g. the singular of a substantive is present in a glossary whereas the plural is missing.
However, there are currently issues with the priority with which glossaries are handled: this conflicts with the new stemming (fuzzy glossary term matching).
Instead of describing the issues, I'd like to present a chart showing the way terms should be searched in several glossaries with different priorities: first search exact, stop if found, if not found: search for fuzzy matching: