The docs for list_to_ord_set/2 don’t tell you that it also removes duplicates. If that’s inherent to the definition of a set in SWI-Prolog then adding a note that indicates that might be considered redundant. But for someone coming to the page that doesn’t know that about SWI-Prolog sets it might be helpful. My two cents:
I can’t think of any programming language or library where a “set” has duplicates. If you allow duplicates, it’s called a “bag” or “multiset”.
See also list_to_set/2.