Re: [web-annotation] Do we need an `annotates` relationship for use in RDF and/or Link Relationships?

Exactly as Ivan says. What we already have is very interchangeable with
"body annotates target". It's one thing that SPARQL can do pretty
reasonably. I expect that other implementations of reasoners will have
similar rule building engines.

The RDF knowledge graph is not a document format. And while Randall's
interpretation of the graph is an idiosyncratic (in the completely
nonpejorative sense), I think it's perfectly fair if he has an annotation
consuming service that simply walks the Web Anno graph into a local
annotation graph (at the price of blowing away all of that sweet, sweet
provenance metadata or keeping a now-redundant set of structured
relationships). Our focus is on the interop and interchange, how
implementations work down at the local levels (inside of repositories,
etc.) is out of scope for us.

The type of search optimizations being discussed are really only useful at
those local levels of implementation, the parts that have nothing to do
with the web itself.

Regards,

Jacob


On Wed, Nov 4, 2015 at 3:16 PM, Ivan Herman via GitHub <sysbot+gh@w3.org>
wrote:

>
> > On 5 Nov 2015, at 05:09, BigBlueHat <notifications@github.com>
> wrote:
> >
> > We can currently express relationships (even vague ones) within an
> Annotation (see #98 (comment)
> <https://github.com/w3c/web-annotation/issues/98#issuecomment-153831097>
>  ).
> >
> > However, we still do not have (for better or worse) the ability to
> state body annotates target. We have instead annotation hasBody body;
> annotation hasTarget target.
> >
> > Should we define an annotates relationship?
> >
> >
>
> Well… if we are in RDF land for a moment, then I do not see what this
> is necessary. That type of relationship between a body and a target
> can be found out by a fairly trivial SPARQL query; using SPARQL as
> some sort of a rule engine (using SPARQL CONSTRUCT) one can generate
> new graphs with this relations, etc.
>
> RDF people have powerful tools already, we do not have to add
> additional things. Non-RDF people will not care…
>
> > Or (perhaps) a link relationship that could be used with either an
> annotation or a body?
> >
> > GET /blog-post/comment-1
> > HTTP/1.1 200 OK
> > Link: </blog-post>; rel="annotates"
> > That scenario done now would look like:
> >
> > GET /blog-post/comment-1
> > HTTP/1.1 200 OK
> > Link: </blog-post>; rel="http://www.w3.org/ns/oa#hasTarget"
> > Obviously, if we specified annotates in the link relation registry
> <http://www.iana.org/assignments/link-relations/link-relations.xhtml>,
>  we'd also (likely) want to specify the reverse relationship for us in
>  the more common scenario of linking from a blog post to any known
> comments.
> >
> > Here are the existing link relationship values that come pretty
> close (but are more specific):
> >
> > bookmark <http://www.w3.org/TR/html5/links.html#link-type-bookmark>
> -- specific to bookmarking
> > describes <http://tools.ietf.org/html/rfc6892> -- description only
> > describedby
> <http://www.w3.org/TR/2009/REC-powder-dr-20090901/#semlink> -- same as
>  above; just points the other way
> > replies <http://tools.ietf.org/html/rfc4685#section-4> -- would only
>  work for a direct reply
> > and...historically (though not part of the registry...yet?)
> >
> > annotation <http://tools.ietf.org/html/draft-ietf-iiir-html-00> --
> ...you'll have to search for it...or see it highlighted
> <https://via.hypothes.is/tools.ietf.org/html/draft-ietf-iiir-html-00>
> > —
> > Reply to this email directly or view it on GitHub
> <https://github.com/w3c/web-annotation/issues/101>.
> >
>
>
>
> --
> GitHub Notif of comment by iherman
> See
> https://github.com/w3c/web-annotation/issues/101#issuecomment-153867813
>
>