Re: Arrays in SKOS: proposal

Miles, AJ (Alistair)  writes:

> Having reviewed all of the discussion on the issue of arrays and node
> labels, I would like to go with the following addition to the SKOS
> schema:
<snip>

Looks good to me.

Following are some off-the-cuff impressions that may or may not be of
interest.

--

A stylistic point: When dealing with boolean properties, I find that
classes are often more convenient.

That is, rather than saying,

    _:1 rdf:type skos:Array.
    _:1 skos:ordered "false"^^xsd"boolean.
    _:1 skos:members ...

it might be better to say

    _:1 rdf:type skos:UnorderedArray.
    _:1 skos:members ...

and have skos:UnorderedArray be a subclass of skos:Array. (The sense
being "this is a kind of array where it doesn't matter if the order of
members is changed.)

--

On the subject of unordered arrays: From a user standpoint, what do you
see as the difference between an ordered array and an unordered array?
After all, even an unordered array must be presented to the user in some
sort of order.

(One possibility: a UI, given an unordered array in a multilingual
thesaurus, might sort items lexically according to the user's preferred
language.)

--

Also on the subject of unordered arrays: I see the advantage to using
the same syntax for ordered and unordered arrays, but there is also one
disadvantage. Consider these arrays:

    _:2 a skos:Array
      ; skos:ordered "false"^^xsd:boolean
      ; skos:members ( A B C D )
      .
    
    _:3 a skos:Array
      ; skos:ordered "false"^^xsd:boolean
      ; skos:members ( B D A C )
      .

At the SKOS level of interpretation, _:2 and _:3 represent the same
array. However, there is presently no way to indicate that at the RDF
level. That is, there's nothing you can put in the schema that a generic
RDF tool could use to infer that _:2 and _:3 are identical.

I don't consider this a problem, but it's worth noting.

--

On the subject of nested arrays: The discussion of this got a bit too
technical for me to follow, but I didn't see anything that convinced me
that nested arrays are necessary. On the other hand, I don't see a great
need to forbid them.
-- 
David Menendez <zednenem@psualum.com> <http://www.eyrie.org/~zednenem/>

Received on Friday, 14 May 2004 01:08:35 UTC