Review Semantic Integration Note

 

Dear all

 

Sorry for the delay of my review. In general it is a very interesting work.

 

Here are some specific comments:

 

Page 2: The third use case is numbered as 1. Why?

Moreover, in Page 3, it is not clear which are the use cases 2 and 3.

 

Page 3: What do you mean by vocabulary? Is it the build-in vocabulary of

an ontology language, say the rdf, rdfs etc vocabulary

or it is the vocabulary defined by an ontology engineer,

like for example the difference between "car" and "auto"?

 

 

Page 6: Principle 4. The text focuses more on reuse and its benefits on
reducing

the work on recreation rather than the interoperability issue.

And it is not clear how reuse would help interoperability.

 

Page 7: "somethigne" -> something

 

Page 7: "along with being a subclass of a new restriction class". This last
comment is not clear.

 

Same page: "The OWL constructs for this are equivalentClass,...". This is

repeated later in the text. Maybe it should be removed.

 

Same page: "The core OWL mapping constructs...". I think that
rdfs:subClassOf

is missing. Maybe this is the only RDFS construct that can be used

for achieving semantic interoperability and since RDF is also covered (the
title says "Using RDF and

OWL"), it is important that this feature of RDF should be clearly pointed
out.

 

Page 8: "OntB:landVehicle". Where is landVehicle in the XML that follows?

Is the concept OntB:landVehicle missing from the XML example?

 

Same Page: Paragraph starting with "Note that although in principle,
any...".

It is quite hard to understand without the help of the comments that follow.

It is better to rewrite it in order to clarify it. 

 

 

Same Page, last 3 lines: "five core OWL...". Are there five or six
(including

subClassOf)?. 

 

Page 13: "yo ucant" -> "you can't"

 

Page 14: I am not really sure for the title "Limitations on OWL". The
following

limitations do not seem like limitations of OWL:

(i) OWL is not responsible for incorrect modeling,

(ii) in any language the user has to provide the mappings...

(iii) the 3rd limitation seems like a software limitation.

 

 

The main comment that I personally have for the document is that, in my
opinion, it does not put emphasis on the advantages and disadvantages of
using OWL (or RDF) for defining mappings. There's a lot of discussion on
this issue in the literature. Several limitations of OWL for this specific
use have been mentioned and lately rule languages have been proposed as more
appropriate for defining mappings. Since W3C RIF WG is running and very soon
a new rule language standard will be the case, maybe it is a good idea to
say a few words for this issue in the Semantic Integration Note.

 

 

I hope this review is helpful

 

Regards,

Giorgos 

 

 

 

 

 

Received on Monday, 3 April 2006 09:05:15 UTC