Re: Identification vs Definition RE: candidate and deprecated concepts

>
>
> Leonard
>
>> I think it would be difficult to publish a set a concepts "in the
>> abstract" without any implicit relationships between them, whether you
>> call this a "scheme" or not. The problem is that the usual way of
>> defining a concept is to say what broader concept it is a member of, and
>> then specify the ways in which it is differentiated from other members
>> of that broader concept.

This is a common (probably as Leonard said the most common) but not the
only way of defining things. A recent use case I had was dealing with
information I needed for a legal case. Since I started with very partial
information I began by building random elements I knew of into my
thesaurus. Bit by bit I started to relate them.

> Main aspects of this issue are:
>
> 1. How do I make distinct in a concept scheme the "defining properties"
> that one cannot
> remove without changing somehow the concept, from other "added
> properties"?

I don't think you can. Once you have described a concept in terms of a
number of attributes, taking away one of those attributes changes the
concept. Whether it is an important change or a minor change is a factor
of what someone is doing with the concept, not of the definition itself.

> 2. If a concept is *identified* by a URI in a concept scheme, is it
> correct to say that it is *defined* by this same URI providing
> you can get through it some non-ambiguous
> information resource about the concept defining properties (subject
> identifier vs subject indicator again ...)

I don't think so. It's a very detailed argument, but I think the concept
identified by the URI is defined by the descriptions that you actually
get.

> 3. If a concept identifier is re-used in another scheme, what should be
> the requirements/recommendations concerning the commitment of
> the re-user to the initial definition of the concept?

I don't think there are any real requirements. But I think there are some
important recommendations related to etiquette and the structure of the
web:

1. Avoid further describing the object of the identifier without some
agreement...

This seems very vague. But for example there is nothing that stops me writing

<skos:Concept><skos:prefLabel>Banana</skos:prefLabel><map:exactMatch><skos:Concept><skos:prefLabel>Washing
Machine</skos:prefLabel></...

(The Skos concepts for washing machines and bananas are the same thing).
But it is not helpful to the world. On the other hand there are occasions
when it is useful to do this - for example a Concept that was
half-labelled by a company that since went out of business, but has left
their thesaurus in some kind of long-term accessible state.

2. Check the persistence and change policies before re-using something

This should be obvious. But different content producers have different
ways of defining things, and have different ideas about what is an
important change to something. It is better to ensure before you use
something that it has a persistence and change policy that isn't going to
cause problems later. (Note that some change policy is generally a good
thing - the number of perfectly defined thesauri which never need changing
to remain applicable to the real world is probably vanishingly small...)

Just some thoughts.

Cheers

chaals

Received on Tuesday, 12 October 2004 04:22:24 UTC