W3C home > Mailing lists > Public > w3c-wai-er-ig@w3.org > February 2001

Re: EARL: Simple Sample

From: Sean B. Palmer <sean@mysterylights.com>
Date: Thu, 15 Feb 2001 00:42:35 -0000
Message-ID: <00fc01c096e8$663b1120$bf6c073e@z5n9x1>
To: "Wendy A Chisholm" <wendy@w3.org>
Cc: <w3c-wai-er-ig@w3.org>
> What do you envision?  An evaluation report of site A written in
> EARL?

Seems as good as any. I think that as EARL has such a wide scope (how many
types of evaluations are there? and there's certainly no end of things to
evaluate), it would do us well to get some examples out. Designing a
langauge for such a wide scope is (IMHO) the most difficult part. Any
evaluation is fine as long as it is a "real life" example, i.e. something
that EARL may actually be used for.

> First attempt will be to assert that I think the alt-text on the image
> that linearizes the table is ok.

Seems very "real life".

> <h:img[@src="/Icons/detab"]> has <h:@alt> of "change column layout"
> <earl:person> <#name> "wendy"
> "wendy" <earl:asserts> "the alt attribute of <h:img[@src="/Icons/detab"]>
> has <l:quality>"

This is good and raises a few points about EARL and representation therof
immediately.

1. The semEnglish method of delimiting URIs with "[]" is flawed for URIs
incorporating XPath.
2. Using namespace aliases to save space writing out URIs... I'm not sure
if this is technically "politically correct", but it's a very good idea. I
think I did it before myself, when I converted one of the W3C's RSS feeds
into N3 for Aaron Swartz.
3. Wendy comments the following:-

> #The assertion shouldn't be a string, but wasn't sure how to fit it into
> the vocabulary or syntax.  All of these should be strung together into
one
> assertion, shouldn't they?

In short, yes, they should. What's happened here is that writing out
triples in Notation3 exposes how close it is to things we encounter in
natural language. Wendy writes the following:-

> "wendy" <earl:asserts> "the alt attribute of <h:img[@src="/Icons/detab"]>
> has <l:quality>"

Which can probably be represented as:-

     "wendy" earl:asserts

     h:img[@src="/Icons/detab"] h:@alt [ l:quality "low" ] .
     } .

N.B. No need to put <> around qualified URIs... at first I thought this
might actually be an improvement on the current, but then I realised that
the <> were there to indicate that this is a URI and not a QName - i.e. is
http:X a URL or a URI with the namespace prefix of http:, and likewise, is
earl: some new URI scheme, or just a QName? Answer: if it's in <> then it's
a URL, otherwise it's a QName prefixed attribute. There have been many
disagreements on RDF IG about using <> to delimit URIs... (even TimBL
joined in on the discussion).

Still, point I'm trying to make in amidst all of my digressions is that it
is difficult to compress things that we want to express in triples, because
we're so used to "natural language". Seth wanted semEnglish to be a step
closer to representing things naturally, and he has suceeded on many
accounts. I think that anyone following this will like to look at the
grammar http://robustai.net/mentography/semenglish.html but beware that
there are still some syntactical errors that Seth needs to fix in the
meantime.

Another thing to look at for EARL syntax purposes might be MetaLog
http://www.w3.org/RDF/Metalog/ but note that N3 was a step away from this,
and semEnglish even further in that the syntax is geting neater, and yet
more human readable. MetaLog is probably the ancestor of these langauges,
and hence has historical importance.

> Not sure what to do about reading order.  Sean, is this the type of
> exercise you were hoping we would do?

Yes - thank you! The more times we attempt to represent "EARL" examples,
the closer we come to a syntax agreement.

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://webns.net/roughterms/> .
[ :name "Sean B. Palmer" ] :hasHomepage <http://infomesh.net/sbp/> .
Received on Wednesday, 14 February 2001 19:42:12 GMT

This archive was generated by hypermail 2.2.0 + w3c-0.30 : Thursday, 9 June 2005 12:10:38 GMT