Re: HTML as the visible part of the Web [Was: files or database for solid pods]

čt 2. 11. 2023 v 10:32 odesílatel Nicolas Chauvat <
nicolas.chauvat@logilab.fr> napsal:

> Hi,
>
> Le Wed, Nov 01, 2023 at 09:20:37AM -0700, Kurt Cagle a écrit :
> > Also, the HTML web is not the Linked Data web. If you assume that you can
> > create a very simplified graph by following <A href> links in web
> > documents, then you could make the argument that the HTML web is in fact
> a
> > subgraph of the Linked Data space.
>
> +1
>
> That's something I keep repeating when I train people to the (semantic)
> web.
>
> 1. There is only one Web to link all the URLs
>
> 2. Part of the data on the Web is HTML documents, but that's only the
>    tip of the iceberg
>

This is true.  There is quite a lot of data in HTML though.  Stats on how
much would be interesting.

Fair to say that both models can live side by side, and benefit from each
other, ie data mime types, html with data

I like also your slide about html views of data


>
> One thing I think we need is the browser to be able to display the
> part of the iceberg that's underwater.
>
> Here is one attempt
>
> https://archive.fosdem.org/2019/schedule/event/collab_cwldbe/attachments/slides/3347/export/events/attachments/collab_cwldbe/slides/3347/nchauvat_cubicweb_ld_browser.pdf
> and here is another one
> https://open-source.pages.logilab.fr/SemWeb/sparqlexplorer/ (yes, it
> needs work)
>

> I need to think about how this can be positioned with respect to Solid.
>
> Comments ?
>

Cool stuff, was the idea to standardize rel="alternative" or ways to pull
RDF into html?

I think the consensus for solid lite would be that full RDF as turtle and
json-ld would be supported by all servers.  This is not hard as as .ttl and
.jsonld are registered with IANA and already in alot of libraries and
tooling.  Similarly, .json and .html will be supported, which should yield
alot of compatibility and interop.

Whether content negotiation would be a MUST or an extension is up for
discussion.  I'll note that content negotiation was added to solid (0.7 ->
0.8) with very little discussion at all.  It might well have been a misstep.

Similarly HTML linking to RDF or other data should be easy enough in both
Solid and Solid Lite, leading to a full web of data, and web operating
system with different devX and learning curves.


>
> --
> Nicolas Chauvat
>
> logilab.fr - services en informatique scientifique et gestion de
> connaissances
>
>