Experience from countries vs capitals

[bvssvni]

I find that Avatar Logic is capable of capturing nuances that otherwise might be hard to reason about:

• Most countries have a distinct city as capital
• Some countries have themselves as capital (e.g. Luxembourg)
• Some countries have multiple capitals (e.g. South-Africa)

When a country has a capital, the capital has the country, but by using pairs one can automatically get "is" when possible in addition to "has":

(X, Y) :- (Y, X), Y : country, X : capital.

When a distinct city is a capital, I use the role:

<city> : capital

When a country has itself as capital:

uniq capital
capital'(<city>) : capital

This does not collide with capital as a role, because capital(x) does not imply a unique value. It is the pair (x, capital'(y)) which implies a unique value.

When a country has multiple capitals, I use another non-unique 1-avatar capitals:

capitals'(<city>) : capital

There is also a country (Switzerland) that claims to have no capital, but which has a "de facto" capital. One can use a unique 1-avatar de_facto_capital'(<capital>) such that capital(<country>) = de_facto_capital'(<capital>) but not capital(<country>) = <capital>. However, (<capital>, <country>) works by lifting out of the 1-avatar:

uniq de_facto_capital
(X, Y) :- (de_facto_capital'(X), Y), Y : country.

When two countries claim the same capital, one can use a non-unique 1-avatar territory'(<country>) with a lifting of the equality, not the pair, to make capital(<country>) substitute with the capital:

capital(X) = Y :- capital(territory'(X)) = Y.