Re: [TURTLE] Turtle Inverse Properties

On 8/20/12 11:37 AM, Steve Harris wrote:
> On 2012-08-20, at 14:20, Kingsley Idehen wrote:
>> On 8/20/12 8:43 AM, Steve Harris wrote:
>>> Just because I don't write Turtle by hand very often doesn't mean other people shouldn't.
>>> Though, to be honest I think we (RDF users) either are past that point, or should be. Writing HTML by hand went out of fashion rather quickly when the web took off, but it was critical to getting it there. IMHO.
>>> - Steve
>> We have to teach RDF by encouraging folks to craft Turtle by hand, as a first step. Making triples visible is the key to
> No we don't. I've taught RDF to a pretty significant number of people now, and I (now) start with a SPARQL database, with pre-written data in it.

We do, at least for the audience profile I have in mind. This audience 
won't be starting with SPARQL at all. They have to understand how to use 
Web documents (type: RDF) to construct linked data resources (content) 
first. Once that's mastered, crawling, indexing, and querying become 
much easier to understand and appreciate.

> Making triples visible is important though.


>> this endeavor. Historically, as exemplified by RDF/XML, losing the triple in syntax ultimately loses the plot. IMHO., HTML with RDFa or Micordata embeded don't address this fundamental issue, neither does JSON-LD (which is for JS developers).
>> The value of TimBL's point is best appreciated once there's acceptance of the notion that folks (profile: end-user and/or integrator / tech plumber) will ultimately start the Linked Data journey by crafting Turtle by hand.
> Perhaps - that's not really an argument in favour of is  of though, that's an argument for making it as simple to read/write as possible IMHO.

Sorta, but his suggestion has useful effect as users become more adept 
with Turtle.
> Reversing the direction of a triple is definitely an advanced requirement, and not likely to make Turtle easier for novices to read.

Correct, which is why I said it shouldn't be at the front-door, at the 
same time, it should be part of the house. Best to have this critical 
part of the house knowing that engaged and enthusiastic users make 
quantum leaps from exercisers of basic functionality to edge case 
maniacs. I am extremely familiar with this end-user transitional 
behavior, from field experience.

> As an evidence point, look at the confusion around @rev in RDFa. e.g.

Unfortunately RDFa and an other attempt to stuff triples in HTML aren't 
great examples. If anything, they ultimately make the case for what's 
being sought in Turtle, at least from my vantage point.

>> Unlike HTML, crafting Turtle by hand is both useful and extremely practical.
> I don't think there's any difference in the utility.

There's a massive difference. It isn't always obvious, but its 
absolutely massive.

>   I handcraft about as much HTML as Turtle.

So do I, but we aren't great test cases at all. Think more about a user 
encountering all of this for the first time. Simple challenge: make a 
webby profile document about yourself. Put in as little as possible 
knowing there's a Web out there that will handle the REST etc..

>   My CV is maintained in HTML (with RDFa in it actually) for example.

Not good examples though, as per my comments and our profiles. We aren't 
the target case. Think about the absolute shortest path to Linked Data 
comprehension and euphoria. In this scenario there's absolutely no 
comparison between HTML and Turtle. Turtle wins flat out, all the time.

Believe me, I understand where you are coming from, but as is the case 
with many aspects of the Web the "deceptive simplicity" of its 
architecture can sometimes obscure a lot of its magic, at first blush.
> - Steve



Kingsley Idehen	
Founder & CEO
OpenLink Software
Company Web:
Personal Weblog:
Twitter/ handle: @kidehen
Google+ Profile:
LinkedIn Profile:

Received on Monday, 20 August 2012 16:52:59 UTC