ICRA Response to Call for Review: Evaluation and Report Language (EARL) 1.0 Schema Working Draft]

Hi Shadi and all,

You invited WCL-XG to comment specifically on your public working draft. 
The XG is currently quiescent while we wait for our rebirth as a WG so I 
can only respond as ICRA, not as WCL-XG.

In section 1.2 [1] you raise the issue of namespaces. I don't think you 
need to issue a new namespace every time you make a change as long as it 
is clear that it's a working draft. After that, of course, the next 
iteration will need a new namespace.

In the core vocabulary, you define earl:Subject [2] as a single 'thing'. 
The approach outlined in the WCL Report [3] will, I hope, be useful in 
future where an Assertor wishes to make assertions about a thing that is 
group of resources. The approach to be confirmed and encoded under the 
proposed WCL WG is flexible and provides a powerful set of methods for 
unambiguously defining such a group.

In section 2.2 you have used foaf:Person, based on this term's 
stability. ICRA (and WCL) uses the foaf:Organization term believing it 
to be sufficiently stable to use. We do so, however, with caution, 
making sure that we talk about the organisation whose homepage is at ... 
etc. rather than giving the domain name _as the foaf:Organization_. I 
would prefer ERT to use foaf:Organization for Assertors that are 
companies rather then individuals.

I really like the CompoundAssertor term!

In example 7, where the Test Subject is a Java applet, you give a 
dc:description etc. and say that it is "part of http://example.org".
I think this might be wrong, depending what you mean by it - and the 
fact that I can't tell what you mean by it suggests that it at least 
needs some clarification.

Do you mean that the Java applet is embedded in the resource that is 
resolved from http://example.org? If so, that's OK (but please clarify); 
but I suspect you mean that the Java applet is part of the website that 
has a homepage at http://example.org?? This comes back to the resource 
grouping issue. Whatever the object of a dct:isPartOf predicate is, it 
should be something that is able to have parts, i.e. a set definition of 
some kind. http://example.org does not obviously fall into this category.

I may be alone in this but I find the term validity level 
(earl:validity) to be confusing. What you mean is "this is the result" 
and the answer is pass/fail/don't know. Validity is surely a particular 
kind of check? As the Class here is called Result, maybe you can use a 
synonym - like 'outcome' rather than validity?

When I first read it, I though validity level meant something like "the 
validity of the test carried out" - something like a degree of 
confidence in the outcome - but no, that concept comes later.  So, in 
concrete terms, please consider using 'earl:outcome' instead of 
earl:validity.

In section 2.6.3 you give vocabulary terms like earl:xPath, lineCharLen 
etc. These seem application-specific to me and I suggest you consider 
splitting these out into a specific vocabulary to be used when using 
EARL to assess Web content for accessibility.

Finally, in 2.8 you have an editor's note about terms for a URI. WCL is 
going to need an RDF vocabulary for the different parts of a URI for the 
grouping mechanism so we have another common interest there. DanBri had 
an exchange with TBL and Dan C on the issue and I think there's probably 
a case for 'someone' just creating a pretty simple RDF vocabulary 
covering scheme, authority, port, host etc.

Hope this helps.

Phil.

[1] http://www.w3.org/TR/EARL10-Schema/#namespaces
[2] http://www.w3.org/TR/EARL10-Schema/#assertion
[3] http://www.w3.org/2005/Incubator/wcl/XGR-report/#scope


-- 
Phil Archer
Chief Technical Officer, ICRA
w. http://www.icra.org/people/philarcher/

Working for a Safer Internet

Received on Monday, 16 October 2006 11:27:56 UTC