Re: ISSUE: Classes as instances

Peter F. Patel-Schneider wrote:
> From: Guus Schreiber <schreiber@swi.psy.uva.nl>
> Subject: Re: ISSUE: Classes as instances
> Date: Fri, 12 Jul 2002 15:12:46 +0200
> 
> [...]
> 
> 
>>Encl. is a detailed example of Raphael's scenario 1.
> 
> 
> [...]
> 
> 
>>One disadvantage of the above representation is that the WordNet
>>hyponym hierarchy is now "hidden" in the hyponymOf triples. 
> 
> 
> Why are you saying that the hyponym hierarchy is hidden?  Do you mean that
> it can't be found?  Surely not.  Do you mean that some aspects of the
> hyponym hierarchy are not captured in this RDFS representation?  Probably.
> Well, so what?  Representation is at least partly about determining what to
> capture and what not to capture.

Assume for a moment that from a certain application perspective one 
wants to use the WordNet hierarchy as a subclass hierarchy. From this 
perspective the subclass hierarchy is not explicit, but "hidden" in the 
instances of the hyponym property.

I am not going to argue here whether the Wordnet hierarchy is 
semantically equivalent to a OWL/RDFS subclass hierarchy, but it is 
certainly not unreasonable to take this viewpoint. Quote from Miller [1, 
p. 25]:

   "[the hyponym relation] isthe transitive, assymmetric semantic 
relation that can be read as as 'is-a' or 'is-a-kind-of'"

>>One could
>>therefore argue that this representation is wrong. 
> 
> 
> Please explain.  What is ``wrong'' about this representation?
> 

I did not say it was (I actually think it's fine). See previous point.
However, I am not sure you would be happy with it, as it impleas that 
WordNet synsets such as denoted by the terms "animal" and "human" are 
represnted as RDF/OWL individuals.

>>However, if the
>>semantic web really becomes a reality, different representation
>>choices will be a fact of life. So, we looked at a way of
>>defining our interpretation of "WordNet as a class hierarchy" as an
>>add-on RDF Schema. 
> 
> 
>>It turned out that with two definitions we could
>>solve this problem:
>>
>>  wn:LexicalConcept rdfs:subClassOf rdfs:Class
>>  wn:hyponymOf rdfs:subPropertyOf rdfs:subClassOf
>>
>>The first definition makes it possible to treat a synset as a class,
>>which is a requirement to able to treat hyponymOf as a sort of
>>subclass relation (i.e. the second definition).
> 
> 
> Well, sure, *if* you want to treat hyponymOf ``as a sort of subclass
> relation''.  However, why would you want to do this?  What do you achieve
> by this?  
 >
> Perhaps the only reason for making hyponymOf a subclass relationship
> derives from limitations in RDFS.   For example, would it not be better to
> make hyponymOf a transitive relation, as is possible in OWL?  

The hyponym relation carries a lot more meaning that just "transitive" 
Anyway, I am not arguing here for the "correct" representation (I do not 
think there is one and only one), just for facilities to map between 
viewpoints.

> [...]
> 
> 
>>With these four definitions the RDF parser and browser used by our semantic
>>annotation & search tool is perfectly happy. 
> 
> 
> Why would the ``RDF parser and browser'' be unhappy without these
> four *triples*?  Are you saying that without these four triples your RDF
> parser and browser cannot read in the documents?  If so, why?

Of course our parser reads in the complete file. What I meant was that
our RDF browser constructs a subclass tree, and does now do this as well 
for the hhyponym relation. The tree is used for annotation & search 
purposes (see below).

> 
>>The tool uses the
>>hierarchy in several ways. e.g.:
>>- for making it easy for users to understand the intended
>>  meaning of a term (e.g. "Venus" as a subclass of "Roman deity" or as a
>>  subclass of "planet"). This is used in annotation/search term
>>  disambiguation.
>>- for query generalization/specialization.
>>[See [3] for more info]
> 
> 
> Why does the tool depend on subclass?  Would it not be just as easy
> to use a transitive relationship?

I think many RDF (and OWL) tools and applications will be treating the 
subclass relation in a special way. What is wrong with that?

>>Although it first seemed like a hack, on second thought this might
>>actually be a decent way to do this kind of ontology/representation
>>mapping. Again, in the semantic web we will have to live with
>>representation choices made by others.
> 
> 
> Agreed, but this does not imply to me that we have to allow classes as
> instances.
> 
> 
>>3. Implications for OWL
>>
>>OWL should not disallow this type of mapping.
> 
> 
> I don't think that you have provided convincing evidence of this
> conclusion.
> 

I disagree. I think it is a realistic example of how things will (need 
to) work in practice.

Every time I present an ontology-representation or -mapping problem, the 
standard answer from some seems to be that either the representation or 
the mapping is wrong. But what are you going to do: install a higher SW 
authority to enforce "correct" mappings and representations?

I am afraid that in this way we will not be building a semantic web, but 
a set of semantic islands with high fensen surrounding them....

>>Comments/suggestions are very welcome,
>>
>>Guus
>>
>>[1] http://www.semanticweb.org/library#wordnet
>>[2] http://www.swi.psy.uva.nl/usr/Schreiber/schema/wnclass.rdfs
>>[3] http://www.swi.psy.uva.nl/usr/Schreiber/papers/Schreiber01a.pdf
> 
> 
> Peter F. Patel-Schneider
> Bell Labs Research
> 


-- 
A. Th. Schreiber, SWI, University of Amsterdam,
Home page: http://www.swi.psy.uva.nl/usr/Schreiber/home.html

Received on Wednesday, 17 July 2002 10:03:21 UTC