Ontologies and statements

[jacoscaz]

Hi all. I come to you once again with another RFC thread focused on the following questions:

1. should a WebID Profile Document obtained by dereferencing a WebID include specific statements about the WebID itself and/or the Agent denoted by it?
2. if so, which should these statements be and why? should they use a dedicated ontology or should they use a well-known ontology (foaf, schema, ...)?

The point of RFC threads is to only carry comments tied directly to the original post, limited to one comment per participant, so as to facilitate my work in taking stock of everybody's position and locate consensus. For more context on RFC threads, please refer to https://lists.w3.org/Archives/Public/public-webid/2023Nov/0121.html .

Pinging @melvincarvalho, @kidehen, @jonassmedegaard, @acoburn, @TallTed, @woutermont, @namedgraph, @csarven, @timbl amongst others.

[TallTed]

4. RFC threads would complement regular discussion threads, with each issue being discussed first and made the subject of an RFC second. I think RFC threads would be best managed as issues on GitHub.

Where is the "regular discussion thread" which preceded creation of this "RFC thread"?

Can you point to any other deployed use of "RFC threads" which has brought a group to successful conclusion of, or even to progress on, any work akin to WebID standardization?

Perhaps seeing such an example would help understanding of how they're supposed to help here.

[melvincarvalho]

Regarding dedicated ontologies I can think off the top of my head of, as:Actor, foaf:Agent, schema:Person and there's some stuff in vcard too. In future there might be another Class, for example, facebook have imlemented WebID in the past.

So if one is picked, that might affect how stable the superset spec is. My impression was that it wanted to be something that was done and then not have to be touched for a long time. I'll defer to @webr3 on this though, as to what he actually wanted in the superset spec.

[kidehen]

1. should a WebID Profile Document obtained by dereferencing a WebID include specific statements about the WebID itself and/or the Agent denoted by it?

That question as posed is confusing. A WebID Profile document MUST include statements that describe the Agent named by a WebID.

[kidehen]

2. if so, which should these statements be and why? should they use a dedicated ontology or should they use a well-known ontology (foaf, schema, ...)?

Schema.org is a broadly used ontology. It is even already broadly used in documents that already satisfy the requirements of a WebID Profile Document.

[jacoscaz]

Thank you @kidehen for your feedback. Ok, let me narrow the scope of this RFC.

A WebID Profile document MUST include statements that describe the Agent named by a WebID.

@kidehen and all - is there any specific statement that, in your opinion, should be required for a document to be considered a valid WebID Profile Document? A simple example would be <webid> a foaf:Agent . if using foaf.

Or, is it enough for the document to contain at least one statement about the Agent, regardless of what is actually stated, for the document to qualify as a WebID Profile Document?

[namedgraph]

I think historically it has always been foaf:Agent. Why would you want to change it and break backwards compatibility?

[woutermont]

1. should a WebID Profile Document obtained by dereferencing a WebID include specific statements about the WebID itself and/or the Agent denoted by it?

Absolutely! From a discovery point of view, a WebID Profile Document is just another Web Resource. Let's not throw REST out the door without good reasons, and have a WebID Profile Document at least state what it is or is about. Not only does this improve its semantics (and therefore human and machine understanding of it), but it also allows us to perform the simple validation after dereferencing that we actually retrieved a WebID Profile Document about the WebID we dereferenced, and not (by human or machine error) some other random Web Resource.

2. if so, which should these statements be and why? should they use a dedicated ontology or should they use a well-known ontology (foaf, schema, ...)?

I second @namedgraph in sticking with FOAF terminology, since afaik it is consistently and almost exclusively used in existing WebIDs.

Not sure it should be <...webid...> a foaf:Agent though. I believe the most crucial statement is the one linking the document to the agent: <> foaf:primaryTopic <...webid...>.

[jacoscaz]

Thank you @namedgraph and @woutermont for your feedback, this is super helpful!

[melvincarvalho]

Given recent discussions, and bearing in mind the aim is to get the spec over line, else all our effort has been in vain:

Consider a catch all solution that would please everyone:

1. A webid vocab that lives in the w3c space, adjacent to the cert ontology, under control of the CG
2. The ontology contains the terms that people want to use in existing vocabs (foaf, schema, activitystreams, vcard)
3. Use OWL to subclass where necessary, or to alias, terms in use
4. New terms can be added to the vocab in future with CG consensus, but existing terms cannot be removed

This should be enough to please everyone, give back compatibility, use the terms needed, allow extensions, be future proofed, be able to write a test suite, all different sersialization profiles, form the basis of an existing widely deployed social graph.

I'll leave it to the chair to decide, but if consensus does not coalesce over one ontology. The catch all solution should unblock everyone, indeed, it might be the best option anyway. There has also always been the need to for external profiles to specify a webid predicate, so this may be needed anyway.

[kidehen]

Please take a look at https://github.com/w3c/WebID/issues/21#issuecomment-1875997800.

Then think about the title(s) of the specs in mind.

A WebID vocab isn't required. Clearly titled specs, comprising consistent content, are what's needed i.e., spec documents that don't conflate problematic items (overtly or covertly).

[TallTed]

I'll leave it to the chair to decide

In the W3C, Chairs don't decide much of anything. They facilitate Group processes toward CG/WG/IG/BG/otherGroup decision-making, optimally by achieving Group consensus, falling back to group vote where consensus is demonstrated to be impossible, and only in very rare cases deciding between two options which have roughly-even support-by-vote in the Group.

Even in such closely-divided voting situations, the goal is to find a compromise solution which allows everyone in the group to do what they need to do with the technology being specified, while not blocking anyone else in the group from doing what they need to do.

As @kidehen says, a (new) WebID vocab/ontology is not needed. Most, if not all, of the necessary predicates already exist in other vocabularies (including but not limited to FOAF and schema.org), and adding any missing predicates to these should be easy, small, and fast.

[melvincarvalho]

1. should a WebID Profile Document obtained by dereferencing a WebID include specific statements about the WebID itself and/or the Agent denoted by it?

To reiterate. Yes. And, importantly, it must be testable.

[jacoscaz]

Closed by #60.