W3C home > Mailing lists > Public > semantic-web@w3.org > October 2010

Re: Vocabulary = Questionnaire ?

From: Bob DuCharme <bob@snee.com>
Date: Sun, 24 Oct 2010 11:04:07 -0400
Message-ID: <4CC44AE7.3030003@snee.com>
To: Sandro Hawke <sandro@w3.org>
CC: semantic-web <semantic-web@w3.org>
Hi Sandro,

rq:short could be very useful in applications built around data using a 
particular schema--for example, in the automated creation of forms. This 
could help frame your diction question. I also think that a property to 
store example values would be a really good idea.

I'm a little confused about rq:comment. You mention "two properties for 
linking RDF  properties to text which presents the property as a 
question", but your rq:comment example is not a question. If it's a 
current best practice to include an rdfs:comment with each property 
declaration, would rq:details be a subproperty of that, with people 
encouraged to use that instead in the future, or do you see any 
difference between how that would be used and how rdfs:comment is 
currently used?

Bob


On 10/24/2010 9:18 AM, Sandro Hawke wrote:
> I like to think of RDF vocabularies as questionnaires, where each RDF
> property is a simple question.  I think this approach makes RDF much
> less mysterious, and my hypothesis is that it will allow people without
> much specialized training to understand and even create pretty good
> vocabularies. This seems like a pretty obvious way to approach RDF, but
> I haven't seen other people, software, or documentation using it, nor
> have a I see a metadata vocabulary to support it.
>
> The main thing I'm looking for is two properties for linking RDF
> properties to text which presents the property as a question.  I think
> it's good to have two properties, because usually you want a short form
> of a question, and then some longer explanatory text.
>
> For example, I'm picturing:
>
>     foaf:name rq:short "What is the name of this person or entity?";
>               rq:details """This is the full name, a sequence of
>                          characters by which this entity is generally
> 			known, with the parts (like firstname and
>                          lastname) in the order used for normal
> 			presentation (not sorting order)."""
>
> Additional metadata like example values (with explanations), and
> importance/salience could be nice, too.  My biggest question is about
> diction: "What is its name?" vs "What is the name?" vs "What is the
> full, common name", vs as above.  I think it will take an effort to
> present many different kinds of vocabularies to many different
> populations to understand the best ways to phrase the questions.   (But
> I think any of these options is still pretty good.)
>
> Where properties are questions, classes used for domains are things the
> questions are about, and classes used for ranges constrain the answers
> and lead to more detailed questions about items provided as answers.
>
> So, has anyone made progress in this direction, and I've missed it?
> Alternatively, does anyone have evidence of shortcomings of this
> approach?
>
> To clarify and motivate slightly: my immediate interest is to help
> people in government work with RDF vocabularies.  My guess is they're
> pretty familiar with filling out forms, and sometimes even designing
> them.   I feel like we need to meet them on their own ground, instead
> trying to teach them to use protégé or something.
>
>       -- Sandro
>
Received on Sunday, 24 October 2010 15:04:31 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:41:24 UTC