Re: [Linking-open-data] watchdog.net and LOD best practices

> Even if the description of a "resource" in the URI rfc say
> that it can refers to actual "things", we only have their RDF 
> representations.

The fact that we might only be able to retrieve RDF through the web does not 
mean that the URI http://example.org/eiffel_tower cannot designate a 
physical object. Please elaborate.

> Well, what is a Class and what is an Individual of a Class? The line is 
> gray, and and someone's class will be another person's individual.

Example: An individual of a tower exists at a certain place at a certain 
time (e.g., the Eiffel tower is currently located in Paris). The class of 
towers is 'made up' of a multitude of towers at many times and places. It is 
very easy to distinguish class from individual when looking at physical 
objects. Of course, making such distinctions gets much harder with more 
abstract things such as 'concepts' or 'data objects'. Which, in turn, makes 
it harder to know where to place relations such  as owl:sameAs, or to create 
ontologies that enable automated reasoning and consistency checking. This is 
why I am a bit critical about using such abstractions as the foundations of 
a large, hopefully somewhat coherent Semantic Web. The identity between 
resources named by URIs and shared semantics are the nodes that hold 
everything together. If they are too loose, everything starts to drift 
apart, and meaningful data integration on a large scale becomes impossible.

> There are certainly best practices such as DOLCE and BFO, but these are 
> not sacred books, and different usecases will have an hard time on them.

DOLCE and BFO are foundational ontologies, not practices. The question is: 
should using such foundational ontologies become a practice?

> But since we are in a Web of Representations, there will be errors and 
> inconsistencies (there are on the Web, and there will be on the Semantic 
> Web)...
> And this is normal: my representation of the World, is not the same as 
> your Representation of the World.

Probably true, and this is why 'representations' are not very useful for 
data integration. We need to postulate the existance of an external reality 
before we can actually talk about errors and inconsistencies.

Many of us are mixing up questions of epistemology (how we as persons can 
gain knowledge) with questions of ontology (how things are in reality). I 
emphatically agree with your subjectivist / constructivist approach towards 
epistemology -- the way I perceive the world might be very different from 
the way you perceive the world. However, the differences in our cognitive 
representations while watching the Eiffel tower do not lead to the existance 
of two separate Eiffel towers in reality (ontology).
The human ability to accept the existance of a consensual external reality 
was a key that enabled the development of science and technology, for 
testing hypotheses, developing theories and integrating many small pieces of 
human experience into a coherent, testable and shared view of reality. We 
should not weaken the practical advantages of this ability by a too radical 
subjectivism.

Sorry for the sudden outbreak of ontological discussion :)

Cheers,
Matthias Samwald
Semantic Web Company, Austria // DERI Galway, Ireland
http://www.semantic-web.at/
http://www.deri.ie/

Received on Thursday, 17 April 2008 13:15:11 UTC