Re: Some reflections on the semantics of embedded triples

Dear Antoine,


> In the current semantics, blank nodes that appear in embedded triples 
> are behaving differently from blank nodes outside. This leads to 
> peculiarities of semantics with possibly unforseen consequences.
>
> TL;DR: In short, blank nodes inside embedded triples can be understood 
> as "placeholders", as opposed to existentials when they are outside 
> embedded triples.

I would not say that they are placeholders,  the idea was originally to 
let them be quantified outside of the embedding.

Taking  your example from below,  the question is whether a triple

:mike :says <<_:x :in :house >>.

should mean (in pseudo FOL, I  am aware  that the below has  no actual 
meaning, but I hope you still get the idea)

∃x: says(mike, in(x, house))

"There is someone (or  something) about whom Mike  says that he is in house"

or

says(mike, (∃x: in(x, house)))

"Mike  says that someone (or something) is in house."

so, it is important to decide where exactly the blank node would be  
quantified.

If  we choose the first option,  the blank nodes occurring in the 
"normal" RDF graph correspond to the referred blank nodes with the same 
name. So, if there is  for example a person of which  mike says that he 
is in house, we could  state;

_:x a :Person.
:mike :says <<_:x :in :house >>.

and the two occurrences of _:x would actually refer to the same instance.

In the second interpretation, the blank node would have local scope and 
the two _:x could refer to two different domain elements. The second  
interpretation would reflect the situation that Mike  actually stated  
that there is someone in house. So, if we have  a graph of statements 
from Mike and that  graph contains the triple _:x :in :house.

Both interpretations have  advantages, but we have to choose one of 
them.  I would say that choice should depend of our  intended use of 
RDF*. So far I understood from the discussions that a correspondence 
between a blank node in an embedded triple and a "normal" triple  is  
what we want.  To get that we have to give up the local quantification.

But I guess  we will have the discussion anyway when it comes to the 
discussion of examples containing bank nodes in our next meeting.

Kind regards,
Dörthe

Received on Monday, 30 November 2020 15:23:00 UTC