Re: JSON-LD in HTML

On Mar 17, 2013, at 04:13 , Gavin Carothers <gavin@carothers.name> wrote:

> 
> 
> 
> On Sat, Mar 16, 2013 at 2:08 PM, Gregg Kellogg <gregg@greggkellogg.net> wrote:
> (Retry, as my mailer seemed to have chosen a personal email address to send it out before. Sorry if it comes twice).
> 
> This proposal provides a means for embedding JSON-LD in HTML documents, =
> much in the same way that Turtle can be embedded in HTML documents. This =
> is important for certain constituencies who have shown problems in =
> providing correct RDFa or microdata, but who are familiar with JSON. =
> Embedding JSON-LD in HTML, along with a JSON-LD context, allows fairly =
> straight-forward JSONto be used for specifying metadata within an HTML =
> document.
> 
> The use of the @data-context attribute below is not ideal, as @data-* =
> attributes are intended for use by other specifications, but it seems =
> inappropriate to either re-purpose some existing HTML (or HTML+RDFa) =
> attribute, or to define a new attribute specifically for this narrow =
> case.
> 
> After section 6.8 in JSON-LD Syntax, add the following section:
> 
> Section 6.9 (new/non-normative) Embedding JSON-LD in HTML documents
> 
> HTML script tags can be used to embed blocks of data in documents. =
> JSON-LD can be easily embedded in HTML this way.
> 
> <script type=3D"application/ld+json">
> {
> "@context": {
>   "@vocab": "http://purl.org/vocab/frbr/core#",
>   "@language": "en",
>   "dc": "http://purl.org/dc/terms/",
> },
> "@id": "http://books.example.com/works/45U8QJGZSQKDH8N",
> "@type": "Work",
> "dc:creator": "Wil Wheaton",
> "dc:title": "Just a Geek",
> "realization": [{
>   "@id": "http://books.example.com/products/9780596007683.BOOK",
>   "@type": "Expression";
>   "dc:type": "http://books.example.com/product-types/BOOK"
> }, {
>   "@id": "http://books.example.com/products/9780596802189.EBOOK"
>   "dc:type": "http://books.example.com/product-types/EBOOK"
> }]
> }
> </script>
> 
> JSON-LD content should be placed in a script element with the @type =
> attribute set to application/ld+json. For text/html, text inside of the =
> script tags does not need to be escaped (issue: use CDATA for XHTML?). =
> 
> http://dev.w3.org/html5/html-xhtml-author-guide/#script-and-style No, XHTML for the web can non contain script tags that contain any form of < or & or ]]> or --. JSON does not provide for anyway to escape those that I'm aware of, therefor you basically can't use JSON-LD inside an XHTML document intended for the web. 
> 
> ... everything else sounds great! Well... except... 
> 
> The character encoding of the embedded JSON-LD fragment will match the =
> HTML documents encoding.
> 
> If a processor extracts the JSON-LD content into RDF, it should expand =
> the JSON-LD fragment into an RDF dataset using the algorithm defined in =
> [JSON-LD-Processing] section 10.6 "Convert to RDF Algorithm". If the =
> HTML file contains multiple JSON-LD script tags, or other RDF statements =
> are extracted, the result is the RDF merge of the datasets.
> 
> Other processors implementing this mechanism may choose to expand the =
> JSON-LD and return the expanded JSON-LD output.
> 
> Section 6.9.1 Specifying Remote Context
> 
> In addition to the mechanism defined in section 6.9, if the =
> @data-context attribute is present on the script element, its value is =
> 
> @data-* SHOULD/MUST? not be used for this. 
> 
> "These attributes are not intended for use by software that is independent of the site that uses the attributes." 
> 

Having been part of the initial discussions on this, I have found exactly the same problem. However... at this point I am not sure what the alternatives are. We can do three things:

- introducing our own attribute, eg, @context; but that is a big no-no with the HTML WG, and it would take a loooong time to convince them of the necessity of one (consider the raws that surrounded @longdesc, @role, or the RDFa attributes). These examples worked out but for standard only, whereas this proposal is for an informal feature, much like Turtle in HTML. Bottom line: that has no chance of getting through, meaning that resulting HTML files would be invalid.
- re-using an RDFa attribute, namely @vocab. But that means that validation of such a file would require validating an HTML file with an RDFa flag (eg, when validator.nu is used) which is strange because this feature has nothing to do with RDFa. And, apart from validation, it is conceptually wrong to reuse an attribute that has nothing to do with RDFa but which would introduced _for_ RDFa. So this may not be an option after all
- using the data-* mechanism which, well, has the drawback you mention.

We have not found any other approach:-( Anybody has a bright idea?

Ivan


> Cheers,
> Gavin
> 
> 
> provided to the JSON-LD processor in a manner equivalent to the =
> treatment of the HTTP Link Header with the =
> http://www.w3.org/ns/json-ld#context link relation as specified in =
> section 6.8, except that the inline specified JSON-LD need not have have =
> embedded context information. For example:
> 
> <script type=3D"application/ld+json" data-context=3D"http://schema.org/">
> 


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
FOAF: http://www.ivan-herman.net/foaf.rdf

Received on Sunday, 17 March 2013 10:46:42 UTC