Re: v1.6 release candidate: Roles, various fixes, site navigation improvements

On Jun 18, 2014, at 4:42 PM, Mark Harrison <> wrote:

> On 18 Jun 2014, at 23:04, "Markus Lanthaler" <> wrote:
>> On Wednesday, June 18, 2014 6:47 PM, Aaron Bradley wrote:
>>> This is not a commentary about the relative usefulness of Turtle, but
>>> on the intended audience of the site and the profile of its adopters -
>>> all the points made by Dan Scott.  An optional Turtle tab?  Sure, why
>>> not.  Turtle as the centerpiece of the code examples?  Only if the aim
>>> of doing so is alienating most webmasters.
>>> And I'd use the same logic in arguing against JSON-LD as the default
>>> view.  JSON, of course, is - unlike Turtle - readily understood and
>>> used daily by most developers.  But remembering, to Dan Scott's point,
>>> that is "a collection of schemas that webmasters can use to
>>> markup HTML pages in ways recognized by major search providers," it
>>> makes little sense to emphasize the syntax which is currently only
>>> tangentially recognized by one of the major search providers in a very
>>> limited context.
>> Yeah, that's a good point Aaron. The problem I see with the current "Without Markup" tab is that it provides very little value to web developers which aren't deeply involved in all these discussions. They mostly spend a minute or two to find an example they can copy-paste-adapt into their sites. It's very difficult to see (visually) that Microdata/RDFa really just add a couple of attributes here and there. Look at for example. All I get from the "Without Markup" table is "well, unsurprisingly the example is about a recipe". Then, when I jump to Microdata or RDFa I get snippets that are first of all almost 50% longer. It's not easy to see what was added. And in most cases I as a developer can't just copy-paste that example as the structure on my page probably looks completely differently. So if JSON-LD would already be well-supported by all major search engines, the whole experience for web developers would indeed become much much simpler in my (obviously biased) opinion. But unfortunately we are not there yet.
>> So, what can we do to improve it in the meantime? An idea would perhaps be to not take "Without Markup" that literally. Mark the examples up with HTML but don't annotate them. Then, when switching to another tab, it's much easier to make the connection.
>> --
>> Markus Lanthaler
>> @markuslanthaler
> Hi Markus,
> I share your concerns about the apparent lack of real support for JSON-LD by the major search engines at present - see below.
> I also like the idea of a Turtle / N-Triples tab but if we'd really like to be able to visualise the examples on as graphs, we could even consider proposing a further additional tab that provides a visualisation similar to the D3.js Tree diagram currently provided via for only RDFa markup.
> Currently, we can of course copy and paste the RDFa example into to obtain the visualisation and extract the Turtle representation of bare RDF triples.  
> However, it might be even more convenient if both of those were available within the web pages as two addition tabs for the examples, rather than having to switch to another site.
> Having said that, I'd really like to see the public-facing Google Structured Data Testing Tool extract *any* structured data at all from JSON-LD markup, whether from the JSON-LD examples or the JSON-LD output from Google's Structured Data Markup Helper.  I can't help feeling that this apparent limitation in Google's Structured Data Testing Tool might be severely reducing the confidence of people to try using JSON-LD markup in their web pages if they keep seeing "No data detected", even though JSON-LD is perhaps more convenient for processing by smartphone apps and less brittle than RDFa or Microdata when for example someone else puts a new hyperlink around an existing image <img> element that is already semantically annotated.  
> I think the ideal situation would be if the public-facing Google Structured Data Testing Tool could successfully ingest and extract structured data from any valid Microdata, RDFa and JSON-LD and provide the D3.js Tree diagram visualisation and bare RDF triples (in Turtle of N-Triples), together with the helpful warning messages currently provided by the Structured Data Linter [ ]


If someone would like to contribute a D3.js visualization to the Linter, that would be awesome! (BTW, the Linter is public-domain, as are the tools on which it is based, anyone is welcome to take anything they find useful). My guess is that the RDFa playground D3.js visualizer could plug-in pretty easily.

BTW, the warnings generated by the Linter are basically just extracting RDF triples and reasoning using RDFS and some rules. It could take Turtle too, for example.


> - Mark Harrison

Received on Thursday, 19 June 2014 01:12:13 UTC