This community generally uses threads when replying, see the member handbook for more info (futureofcoding.org/member-handbook#how-we-slack).

I'm not using one. I'm looking for examples to use as inspiration. But I'm familiar with SWIPL, if that's convenient?

Logic languages come in all shapes and sizes, so it's mostly down to how you format the data you're searching over.

If you're using a function/term-style format, like X(Y, Z, W, ...) , you could introduce variables for the term, and do some variadic-like syntax.

Something like X(..., socrates, ...) :- ...

That's for term-structured data.

Yeah, that's what I'm looking at.

Term-structured. That ellipsis syntax could be built, but isn't?

It could. Depends on how you do resolution.

Also depends on the limits you want to impose.

X(..., socrates, ..., plato, ...)

And how deep you want to nest these terms.

Arbitrary depth. I'm not really looking for ideas, I'm looking for examples of where it has been implemented already, if anywhere. Just because I need an LLM to be able to reliably use it, so it will help a lot if it's in the training data already.

I sort of feel like it's not likely to exist as semantic sugar anywhere, because it would be semantic sugar for a really inefficient sort of search. And most Prologs are usually being used for programming, not KR. Just hoping against hope.

I wonder if in ErgoAI you would do something like this... _?x[?y->term];term[?x->?y];?x[term->?y]._