FW: Ontology browsing

FYI

-----Original Message-----
From: Karun Bakshi [mailto:karunb@ai.mit.edu]
Sent: 23 October 2003 18:05
To: 'Butler, Mark'
Subject: RE: Ontology browsing


Hi,
  I suppose it should not be too difficult to develop an ontology diff
tool.  I think we just need to do a graph diff and color the
intersecting and XOR portions appropriately in the display.  The tricky
part would be what part of the ontology's you want to compare, e.g.
compare if two ontologies share the same property, or see if the same
property in both ontologies has the same range.  

Also, speaking of versioning, I have been wanting to develop an ontology
server that you can ask for an ontology with a particular version, and
it ships you the XML representation of the ontology.  I'm not sure how
this fits into your plans, but just thought I'd throw it out to see if
you have thought of it or if there exists a standard protocol for it.  I
have not mentioned it to Prof. Karger yet, but I think we will/should
need it soon.

Karun

-----Original Message-----
From: Butler, Mark [mailto:Mark_Butler@hplb.hpl.hp.com] 
Sent: Thursday, October 23, 2003 12:44 PM
To: SIMILE public list; 'Karun Bakshi'
Subject: Ontology browsing

Hi team

First thanks to Karun for getting back to me quickly. 

I wanted to explain a bit more about the type of ontology browsing I
want to
do. I doubt whether tools do support these use cases, but I think it's
worth
recording them. 

Use Case 1. "What are the differences between these ontologies?"

I'm interested in this one as I want to see the differences between the
VRA
core schema I wrote, the one Andy wrote, and the one I wrote based on
Artstor. However, because the ordering of the property and class
definitions
is arbitary, you can't do this just using diff. In fact, unfortunately
when
I decided to stop editing RDF/XML and switch to N3, unfortunately my
schema
got rearranged which didn't help.

I think a tool here would help, particularly in applications with
versioned
vocabularies (e.g. UAProf).

Use Case 2. "What is the difference between the output of this XSL
stylesheet and this ontology?"

(The following is a brief introduction to RDDL, I mentioned this on the
call
so you may want to skip)

In the future, if RDF is to take off, we are going to have to migrate a
lot
of XML to RDF. As I said on the call, I like the idea of doing this by
XSLT.
I'm sure its not as good as Perl, but I like the idea of using XSLT with
RDDL (Resource Description Discovery Language) e.g. we have a piece of
XML
using a specific namespace, at that namespace we have an RDDL document
that
describes a set of resources e.g.

1. a XSLT stylesheet to style the XML to RDF/XML
2. an XForms document for entering data using this schema
3. an XML schema document for data validation
4. an RDF Schema for the RDF/XML representation
5. an XSLT stylesheet to style to HTML to make it human readable
6. an OWL schema

as you can see, providing these resources allows the data to be used in
many
different ways. Now there is an advantage in using XSLT here rather than
Perl, as no one will want to download Perl on demand as it could contain
viruses, although hopefully XSLT is less dangerous although a rogue
transform could execute an infinite loop and hog resources. So I think
its a
nice aim to demonstrate that XSLT can be used here.

The other nice side-effect is that human author's don't have to write
RDF/XML, which I think is pretty horrible. They can just write XML, then
the
only person he needs to consider the RDF/XML is the stylesheet author as
even when the data is in RDF it can be serialised using a more human
friendly format like N3. 

(end of RDDL description)

So if we accept that it will be a regular occurance to style XML to RDF
via
XSLT, then we need tools to help automate this, in particular check that
our
schemas and XSLT transforms match up correctly. I've been experimenting
with
command line tools to do this as I've already mentioned - I just extract
the
classes and properties written by the stylesheet, and then diff them
against
the properties and classes defined in the schema and check they match
up. 

However this is a complicated process, so it would be nice to have some
automated GUI tools here that helped the user through the process. 

Dr Mark H. Butler
Research Scientist                HP Labs Bristol
mark-h_butler@hp.com
Internet: http://www-uk.hpl.hp.com/people/marbut/

Received on Thursday, 23 October 2003 13:28:45 UTC