Re: [rdfweb-dev] Re: licence for Ontologies

On 27 Oct 2004, at 16:41, Ron Alford wrote:

> Danny Ayers wrote:
>> Regarding Ron's points about alternate view of terms - well yes,
>> someone could mirror say the FOAF spec with additions and
>> modifications and that could be valid according to the RDF model.
>
> I'm not sure what the big deal is about alternate versions of 
> Ontologies.
> If you take the straight foaf ontology and apply it to Live Journal's
> data, you will get tons of false inferences (in the real life sense,
> anyways). They are, by default, using an unpublished and distinct
> version of Foaf.

Should they not then publish in their own namespace  the differences 
they brought to the official foaf? If they created a concept that was 
different enough from the one of the official foaf, would it not be 
right baptize it in their own namespace?

>
> In my own copy of Foaf, I've been tempted to model Organizations as a
> subclass of Group.  This would allow me to do membership and other
> things that are beneficial to what I'm trying to express.

Could you not also create a Group class in your namespace, and then 
state somewhere in your ontology that foaf:Organisations is a 
superclass of ron:Group. But perhaps you are thinking that by putting 
this statement on your web site, then in some sense you are extending 
the foaf ontology....

> Taken together, the axioms in the ontology form rules for expressing 
> the
> implicit.  My use of foaf:Organization is one way of several to express
> that I'm talking about the same thing, but disagreeing on the
> definition.  See wikipedia for some examples of disagreement on
> definitions.  Do we expect the semantic web to be any different?

Part of what is nice about the semantic web is that we can be precise 
about definitions and what we are speaking about.

	This reminds me of David Lewis' philosophy of Language. Let me digress 
a little. David Lewis defined the meaning of a sentence to be the set 
of possible worlds in which it is true, thereby making meaning a very 
precise mathematical abstraction. Natural languages on the other hand, 
as we all so well know, are not as precise as this. His explanation of 
this difference is that a natural language maps to a set of his precise 
mathematical language objects. A word can have many different meanings. 
We are all speaking an infinite number of mathematical languages 
indeterminately. At some point this indeterminacy comes into the focus 
in practical life, and we then make a choice to narrow down the meaning 
of our terms.

	So if I get back to your example of the word "organisation". It is 
true that the english word has many different meanings. The nice thing 
about the semantic web is that you can decide to create your own 
ron:organisation, I can create henry:organisation and Danny can create 
danny:organisation and we could all be surprised to find out that they 
are all equivalent later on. So there is no need for me to take 
ron:organisation and change its meaning.

> My instance data specifically imports this copy of foaf, so that
> reasoners know exactly what rules I'm following, and can classify
> accordingly.

but perhaps you could have a rule file that simply says
if X is a foaf:organisation then it is a ron:organisation
and then keep on working with rules you are developing relating to 
ron:organisation
>
>
>> Things could get complicated if they introduced assertions that led to
>>  inconsistencies with the 'official' schema (in human or logical
>> terms). Say someone wants to say "John knows Rioja", and add the
>> statement:
>> foaf:knows rdfs:range vin:Wine
>
> Does this somehow corrupt the original foaf?  No, because the original
> ontology at it's published and findable location is still intact.  It
> boils down to a problem of contexts, and if you don't keep context in
> mind when working on the web, you're sunk before you even begin.

Yes I agree this is a good point. I am not yet sure how to work with 
contexts. I need to read up on that. Any good pointers?

>
>> Whether anything in human law, copyright or whatever, may be useful in
>> encouraging consistency, I don't know.
>
> Consistency is already heavily tilted towards the owner of the URI.
> However, trying to mandate definitions in RDF seems futile and
> aggravating. to me.  Does foaf need specific permission in order to
> subclass foaf:Person?

This is a very good point. Owning the domain name in which the words 
are expressed gives the owner of the domain name a huge advantage. And 
so perhaps this is the reason why a BSD license is no major problem.


> What if want to express the ontology in RuleML, DAML, First order 
> logic,
> OWL with E-connections, etc?

good points all...

> -Ron
>
> _______________________________________________
> rdfweb-dev mailing list
> rdfweb-dev@vapours.rdfweb.org
> wiki: http://rdfweb.org/topic/FoafProject
> http://rdfweb.org/mailman/listinfo/rdfweb-dev

Received on Wednesday, 27 October 2004 17:40:58 UTC