Re: ISSUE-283 (Jonathan Jeon): a microformat for mobileOK trustmark [mobileOK scheme]

Hi Jonathan ad all, comments in line below.

This by way of a written version of the comments just made during the 
MWBP call. The content of this e-mail is given as a response to the one 
Jonathan posted to the POWDER public list but also replies to similar 
comments he posted to the MWBP list.

전종홍 wrote:
> Dear all, 
> 
> Currently, we (me and korean community) are trying to apply the machine-readble 
> mobileOK trustmark (like POWDER) within our pilot service.

Good - I hope it works for you!

> 
> But, the trustmark of POWDER style seems to be not easy to use on current browser. 
>  
> I think there are some issues to use of POWDER-based trustmarks.
> - not simple, complex

That's a matter of opinion of course. We have tried very hard to make
POWDER as simple to create and use as possible. A Description Resource
has to define the scope (what is described), the description itself and
for all that to be attributed to an identified agent of some kind. Those
3 elements obviously put a lower limit on how simple it can be.

> - cannot apply to current mobile browser

True. Using POWDER directly implies some development work of come kind.
A POWDER Processor is the way we suggest to go from a POWDER doc to
'native RDF', and GRDDL gets you from POWDER to something very close to
what can be understood in OWL reasoners.

> - cannot append within (X)HTML file 

True. This was an early decision. Our use cases, including the mobileOK
one, pointed us to separating the description and the content it
describes. Remember that POWDER is designed to describe lots of content
at once so that you only have to process it once to know that, say, the
whole of example.org is mobileOK. We didn't want it to be necessary to
fetch and parse a resource to find its description - you often want to
know before hand whether you want to bother fetching and parsing a
resource for a particular application or delivery context.

> - should manage separated POWDER xml file 

See above. This was very much a deliberate design choice.

> 
> For this reason, I think we need to consider 
> the simple mobileOK trustmark that like microformat style.
> 
> So, I was starting to define a mobileOK trustmark of microformat style.
> 
> http://docs.google.com/Doc?id=dhpvgnmn_68dwzhntdt    
>  
> It is simple, but also similar and compatible with POWDER. 
> 
> It has some cons ... 
> - simple, not complex

Again, a matter of opinion (i.e. it's simpler if you're a microformats
person, more complex for others)

> - compatible with POWDER

Up to a point in that it uses the same constructs, yes, but it puts an
overhead on processing that I would suggest can be avoided by keeping
the data separate.

> - can use widely available browsers

That's a strong plus point of course.

> - can embed in (X)HTML

Well, as you'll have guessed, I don't personally see that as a positive.
The examples you give in your document add several lines of code to an
XHTML document, code that would need to be repeated in every mobileOK
document - which, surely, goes against one of the general aims of mobile
content provision which is to keep page weight as low as possible?

For this reason, I think an approach closer to the RDFa model would be
better if you want to embed the mobileOK claim in the document. I know
it's about to disappear from the next version of the mobileOK scheme but
take a look at [1]. That example adds in the mOK namespace and uses the
typeof attribute [2] to link to the 'Conformant' vocabulary term.

Now... one thing that this approach loses by not using POWDER, that your
microformat retains, is the attribution. This is critical for POWDER so
I'm glad you've retained it!

Whether the microformat approach is better than POWDER depends what
you're trying to do of course. If you want to include a claim in an
XHTML document that it is mobileOK, then the microformat approach works
- although I think it might make a page not mobileOK in the process.

A more compact mark up would be better - and RDFa does the job nicely
(although I dare say a microformat could do the job just as well). Link 
elements work in all versions of HTML and achieve the same thing. HTTP 
link headers are efficient for all formats, not just HTML.

If you want to declare that a lot of resources have been found to be
mobileOK and that anyone that wants to verify the claim is welcome to do
so by following various links, then I think a separate POWDER file
offers a better approach.

A couple of specifics about your document:

1. It's wrong to suggest that
http://authority.example.org/company.rdf#me describes W3C. The URI is a 
made up 'example' one.

2. In your first POWDER example you've quoted the currently published
version which has the made-up URI for the mOK logo. Please update that
as you have in the other examples.

3. Your Example 3 uses the sha1sum property (element) which comes from
our example where one DR certifies another. It would be wrong to include
a hash as a description of, as you have it, lots of things on
example.com and example.org.

Obviously I and the rest of the WG are delighted that you've taken the
time to study POWDER and it's good to see it being talked about and
promoted. I do, however, think that the re-expression of it as a
microformat risks losing sight of what POWDER was designed to do and
therefore to retain all its features in a new environment may be pushing
it too far. Better to take what you want and live with what you lose on
the way, knowing that if you want the full feature set, especially in
terms of processing efficiency and page weight, you need the real thing.

Cheers

Phil.


[1]
http://www.w3.org/2005/MWI/BPWG/Group/Drafts/mobileOK-Trustmark/20081017#rdfaClaim
[2] http://www.w3.org/TR/rdfa-syntax/#rdfa-attributes

> 
> I think it will be helpful to spread the POWDER based applications. 
> what do you think about it ? 
> 
> Best Regards, 
> 
> --- Jonathan Jeon 


-- 
Phil Archer
w. http://philarcher.org/

Received on Thursday, 6 November 2008 15:59:53 UTC