> Microformats add semantic meaning to individual HTML elements; this is in contrast to document-level semantic formats like Microdata, RDFa, and JSON-LD.
where did this misconception come from? this is not true. json-ld has @id that can refer to specific elements on the page. microdata and rdfa both directly mark up existing html content.
the very links on this adapt wiki to rdfa & microdata have examples showing this:
<button
itemscope
itemtype="https://www.w3.org/ns/aui/"
itemprop="action"
content="undo">Revert</button>
https://github.com/w3c/adapt/wiki/Comparison-of-ways-to-use-vocabulary-in-content#2
there's also itemid... the page can define multiple different resources on it, give a uri to the different bits of content on it.
microdata & rdfa are excellent & wonderful ways to describe individual html elements.
@Seirdy you have very strong convictions that microdata/rdfa doesn't describe anything on the page, but i frankly don't think there's any rules to the effect that we have to disregard & have to consider the annotations to be completely decoupled & have no correspondence to the elements they decorate.
this seems like an incredibly willful blindness that 98% of webdevs, whatever their prior experience, would not believe or understand or think was true or helpful in any way
of course itemtype describes the element. of course itemid describes the elements id. that it is *used* for building rdf triples is true. but that doesn't mean that markup is literally unallowed to be used for any other purpose.
this stance you have is extremely hardline & deeply sabotaging, vastly rejects possibility. and it promotes a kind of grotesque "microformats are the only way" illness. which is self serving. this is a fake win you are chalking up and it's gross.
“microdata and rdfa both directly mark up existing html content.”
@jauntywunderkind420 Only on a syntactical level. Beyond that, it’s the other way around: Microdata and RDFa let you build RDF triples by taking content from HTML elements. They don’t exist to add information to HTML elements.
HTML and Microdata/RDFa syntaxes being mixed into the same document doesn’t change this. They’re only combined so human- and machine-readable data can remain identical without repetition, not to create more linked-data than the sum of its parts. You could re-write the very same RDFa data in e.g. Turtle syntax, with the HTML discarded. That’s why I wasn’t particularly fond of the (rejected) Microdata/RDFa approach to WAI-Adapt. (I think this was one of the major criticism raised, but I can’t seem to find it)
An @id in JSON-LD takes information from, rather than adding information to, the HTML document; HTML-parsing user-agents should ignore it unless they’re extracting RDF data from Microdata/RDFa.
“there’s also itemid…the page can define multiple different resources on it, give a uri to the different bits of content on it.”
itemid is just a way to refer to add to another object instead of creating a new one. Again, it’s a way to extract information from different parts of an HTML document (or different HTML documents altogether) rather than a way to add to HTML elements.
Microformats sometimes use HTML itself rather than extracting data from it (see the e- prefix). This puts microformats in a sort of gray area, extracting data from HTML but also integrating with it.
I make extensive use of Microdata and microformats on seirdy.one. Strictly speaking, both syntaxes are made of HTML; however, I’ve certainly found the microformats approach to feel more “HTML-native” than Microdata.
POSSE note from https://seirdy.one/notes/2022/09/13/rdf-versus-semantic-html/