This week I will do a “one post per day of something I'm excited about” hashtagged #knowledgegraphs.

Today it’s RDF+Surfaces! Initially coined by Pat Hayes but now implemented and primed by @hochstenbach and Jos De Roo (who maintains the EYE reasoner in Prolog).

This has the potential to fundamentally define data “policies”. Whether it's about access control, inferences, retention policies, etc. Surfaces can do it all

Spec: https://josd.github.io/surface/
Exciting demos: https://github.com/phochste/Notation3-By-Example/tree/main/log/blogic

RDF Surfaces Primer

@pietercolpaert,

Does that all breakdown as follows?

[1] Sentence --> a triple
[2] Paragraph --> a collection of triples pegged to a common predicate
[3] Page --> named graph in SPARQL
[4] Blank Node -- indefinite pronoun for denoting the subject of object of a triple

What do you think?

/cc @hochstenbach

@kidehen and the idea that each page can have a “color”, like: everything on a page could be explicitly not true. @hochstenbach

@pietercolpaert ,

Yes.

A page is an observable thing with discernible attributes that can be expressed in descriptive sentences.

Anyway, are we in agreement regarding 1-4?

@hochstenbach

@kidehen @hochstenbach I find it hard to not agree with you since you're often my reference 😅

Not sure how deep you want to go, however:
[1] Statement = triple. A sentence could contain multiple statements at once
[2] What do you mean with pegged around a predicate?
[3] It’s a bit unclear what it means for a triple to be “in” the named graph. That's why we’re thinking N3 and RDF* rather than using named graphs. But general idea: yes

(1/2)

@pietercolpaert ,

" [2] What do you mean with pegged around a predicate?"

A collection of triples associated with a common predicate i.e., a relation.

Basically, that a Paragraph is a Relation -- sticking to this sentences, paragraphs, and pages perspective.

If I have the following RDF sentences:
{
<#this> <#relatedTo> <#that> .
<#this1> <#relatedTo> <#that> .
<#this3> <#relatedTo> <#that> .
}

<#relatedTo> is the relation predicate (or basis).

/cc @hochstenbach

@kidehen @hochstenbach I don't think [2] paragraph is of any use in the RDF+Surfaces world, but I understand where the parallel comes from. If you want to group triples together around a subject, I see more interesting comparisons in TREE or Hydra’s collection design

@pietercolpaert,

Here's a clearer breakdown as I left out some important items initially -- and reorganized the ordered list.

[1] IRIs denote observed things from a worldview

[2] Sentence (a triple comprising a subject->verb->object) -- describes one aspect of said thing

[3] Paragraph --> a description collection (relational set based on a predicate)

[4] Page --> Document (a Named Graph in SPARQL) comprising descriptions

SeeAlso.
https://en.wikipedia.org/wiki/Linguistic_typology

/cc @hochstenbach

Linguistic typology - Wikipedia

@kidehen Mmm, sure... I am just not entirely sure why we are doing the exercise: just for advocating RDF to developers?
@hochstenbach

@pietercolpaert,
It has a lot to do with what #RDF and surfaces upon which triples are inscribed.

The linguistic aspect of RDF, in regards to “parts of speech,” is very helpful to understanding many subtleties that are sometimes lost in confusion.

Ultimately, this is about storytelling for audiences that still remain a little confused about this important standard for structured data representation 😊

@hochstenbach

@kidehen @pietercolpaert I was out for 2 days. In RDF Surfaces: we don't intend to touch/change the semantic meaning of triples/graphs/blank nodes but the logic and modes of transport. On surfaces triples and are not sentences (triples don't have any order, sentences do) and surfaces are not pages in the same sense. A surface provides a border of a graph like a the parentheses in a logical formula (A AND B AND (C AND D)) : 2 nested formulas. In N3 this would be 2 nested formulas/surfaces

@hochstenbach,

Sentences don't have an order, but they have fundamental subject->verb-object structure.

Sentences don't float in space i.e., they are inscribed on a surface.

What did I miss? 😀

/cc @pietercolpaert

@kidehen @pietercolpaert Well, Hamlet on a positive surface is still Hamlet. Hamlet on a negative surface means "To not be, or to be" :)

@hochstenbach,

"An RDF surface is a kind of a sheet of paper on which RDF graphs can be written." (excerpted from https://josd.github.io/surface/).

The paragraph above actually articulates what I am trying describe.

I was just trying to confirm conceptual alignment using everyday terminology.

An #RDF doc is just a digital rendition of a piece of paper 😀


/cc @pietercolpaert

RDF Surfaces Primer

@kidehen @pietercolpaert It is tempting to use 'sentences' and 'paragraphs' and using words like 'quoting' (as done in RDF*) as a natural language. But, when coding data on an RDF Surface..when natural language and logic meet, it is the latter one that makes the rules. On a paper one could write "to be AND not to be", on an RDF Surface an implementation would complain "this is a contradiction!".

@hochstenbach,

Typically, it would be statements and relations regarding what #RDF does without dropping too deep into some of its more esoteric aspects.

#RDF as a powerful (an pretty unique) data definition language is a key theme that isn't broadly understood or accepted just yet.

I want us to collectively change that 😀

/cc @pietercolpaert