Advancing Web Metadata

Semantic Web Interest Group,

Hello. I am excited to share a new WICG proposal, Advancing Web Metadata: https://github.com/WICG/proposals/issues/141 . I hope that this proposal is of some interest to you. Any feedback or ideas towards improving it would be greatly appreciated.

Introduction
One of a number of topics explored in that proposal involves how best to go about encoding metadata about documents' elements both with and without ids or URIs.
A use case scenario is that of document-element-granular provenance and authorship. For real-time document co-authoring scenarios, metadata could be used to detail users' contributions. Document elements, e.g., paragraphs, could have metadata including their contributors. These techniques would be additionally useful for using metadata to describe which paragraphs, if any, were AI-generated.

Attribute-based Metadata
Let us consider a new attribute "msrc" for providing metadata on HTML elements.

<p id="p123" msrc="data:text/n3;base64,QHByZWZpeCBkYzExOiA8aHR0cDovL3B1cmwub3JnL2RjL2VsZW1lbnRzLzEuMS8+IC4NCg0KPGh0dHA6Ly93d3cuZXhhbXBsZS5jb20jcDEyMz4gZGMxMTpjb250cmlidXRvciAiQWxpY2UiIC4=">...</p>

In the above example, the base64-encoded attribute value is:

@prefix dc11: <http://purl.org/dc/elements/1.1/> .

<http://www.example.com#p123> dc11:contributor "Alice" .

In the example, the URI <http://www.example.com#p123> identifies the paragraph element with id "p123", the paragraph bearing the attribute with the encoded metadata.

This technique runs into complexity when the described document element does not have an id and, thus, URI.

<p msrc="data:text/n3;base64,???">...</p>

The document content, in this case a paragraph, might be desired to be portable, along with its metadata, for purposes including sharing, clipboarding, or dragging-and-dropping it into other documents. When that content with its metadata would arrive in a new document, its XML/HTML id might be changed or omitted and its metadata would still be desired to describe it.

As considered, such metadata would require a processing context to replace a placeholder, a special symbol or URI, with a specific pertinent subject URI.

One technique would involve introducing a new symbol, e.g., "^", to serve as a placeholder:

@prefix dc11: <http://purl.org/dc/elements/1.1/> .

^ dc11:contributor "Alice" .

A second technique would involve a special standard URI for serving as a placeholder:

@prefix dc11: <http://purl.org/dc/elements/1.1/> .
@prefix about: <http://w3.org/ns/about#>

about:this dc11:contributor "Alice" .

A third technique would involve providing attribute-based metadata in special content type – a type other than "text/n3" – with a syntax to enable a preprocessing step. One question would be how best to specify the post-preprocessed content type, e.g., "text/n3".

@prefix dc11: <http://purl.org/dc/elements/1.1/> .
@prefix about: <http://w3.org/ns/about#>

<? this(); ?> dc11:contributor "Alice" .

RDFa Metadata
Here are some of the same topics from attribute-based metadata phrased with HTML+RDFa metadata.

In the following example, the RDFa metadata in the paragraph is about the paragraph document element.

<p id="p123" about="#p123">...</p>

The complexity, indicated above, would arise when the document element, in this case a paragraph, did not have an id or URI.

<p about="???">...</p>

As indicated, above, options would include, but not be limited to, a new syntactic symbol as a placeholder:

<p about="^">...</p>

or a special, standard URI:

<p about="about:this">...</p>

Conclusion
There may be other (existing or new) approaches and solutions beyond those brainstormed and broached above. Any feedback would be greatly appreciated with respect to both the shared proposal (https://github.com/WICG/proposals/issues/141) and these topics. Thank you.


Best regards,
Adam Sobieski