> I'm a bit puzzled by a ranking with at preference in an "instance of" but...
It’s indeed an interesting point. The problem in the country domain is that there is a lot of evolution into the regime of a state, a state can be sovereign at some point in history then become a part of a bigger state losing its sovereignty. If we assume a kind of continuity of a state across this status change, we have to use ranks to select the last valid value. There may not be items for all the regime of a state in history, and a practical choice could be to store the information in the « instance of » statements with date qualifiers.
In this case I assume however it’s just a practical way to circumvene the complexity or even inexistence of our ontology on countries. I would in most case would have noted that « sovereign state » is a subclass of « country ». If you want to include countries of all kind and don’t miss any, you’d have to use a construction like
> ?country wdt:P31/wdt:P279* wd:Q6256.
There for example a subclass of « wd:Q6256 » that is « former countries », so this query would include former entities.
My opinion on these is that if we choose a scheme where there are classes or former entities, the best would be to have the counterpart « today’s country » (with label (country) and a superclass for both, « country (either former or not ) ») to avoid having the « former country » class be a subclass of «today’s country».
Another question, why is not « sovereign state » as the sole class not enough for this query ? Or only the state recognized by the United Nations (I don’t know if/how we model this) ?