well, it depends a bit

Your request sounds like filtering, not like splitting, however, both is possible...
there are datfiles out there which do use special region and language settings.....and of course you got 'normal' dats which simply encode region specific attributes in the set name...like "pacman (EUR)" for example...
For such cases you can go to scanner->set information (button in the scan tree window bottom left), then use the "select sets" edit box and hit apply afterwards. This will only enable the sets you select in the edit box (more later). Only activated sets are scanned/rebuilt/merged, so if you limit your set selection, you can e.g. rebuild EUR-only files to a new folder.
Now to the select sets edit box. This box understands some kind of regular expressions and variables... The variables are shown in the tooltip of the drop down box (move the mouse cursor over the arrow). Regarding regular expressions, you can e.g. enter %d=*(World)* to enable all sets which got a "(World)" in their description tag (%d). You can play around with logical not etc....