W3C home > Mailing lists > Public > public-owl-wg@w3.org > September 2008

Re: English examples in the OWL 2 syntax specification

From: Christine Golbreich <cgolbrei@gmail.com>
Date: Tue, 16 Sep 2008 10:07:29 +0200
Message-ID: <b0ed1d660809160107y77dae2d0x843363175eadb49b@mail.gmail.com>
To: "Boris Motik" <boris.motik@comlab.ox.ac.uk>
Cc: "Kaarel Kaljurand" <kaljurand@gmail.com>, public-owl-wg@w3.org

Hi Boris

First of all, including these examples in the spec was really a good
idea.The irregularities (I had noticed it a while ago when looking at
the spec for making the Requirements consistent with it) do not seem
so much  important, compared to the benefit of having such  examples
in the spec.

If you decide finally to fix it, some of them may be easily
uniformized, for example for ClassAssertions, DisjointClasses,
FunctionalProperty, PropertyAssertion etc. :

DisjointClasses( a:Boy a:Girl ):  Nothing can be both a boy and a
girl. -> No individual is both a boy and a girl
DisjointClasses( a:Man a:Woman ) No object can be both a man and a
woman. ->No individual is both a man and a woman.

FunctionalProperty( a:hasFather ) Each person can have at most one
father. -> Each individual has at most one father
FunctionalProperty( a:hasName ) Each object can have at most one
name.-> Each object has at most one name.

etc.

> I'd like to hear from others about what kind of approach to adopt

In general, I would be inclined to favor "a more OWL-centric one"
rather  than a natural language (or ACE) explanation
because:
- people may certainly be able to translate a more OWL-centric into
natural-language by themselves, since the given examples are really
simple common sense examples.
- Users often prefer a more precise explanation, closer to OWL syntax
rather than natural language which is ambiguous (Perhaps quite
surprisingly enough, but they even often preferred the DL syntax
displayed in Protégé than the Manchester syntax in P4!)
- The examples in the OWL Web Ontology Language  Reference were in
general rather "OWL-centric " and it was quite used and appreciated.

In fact, the question of the approach may have more importance only
for some of the less obvious new features, e.g. asymmetry. I woud say
that we may be flexible, and add for these few specific cases a more
informal sentence in NL, if needed.

In contrast, I found more embarrassing the following in the Syntax
about the examples:
in most cases the example that illustrates a primitive/axiom is given
together with another (or a set of other) axiom(s) introduced by
"Consider the ontology consisting of the following axiom". Very often,
this other axiom(s), which appears first and is highlighted by its
position at the beginning of the example, a special style, font and
tab, is NOT the axiom corresponding to the presented construct!

e.g., for  DataSomeValuesFrom in 8.4.1 Existential Quantification, the
highlighted axiom that we see first is:  PropertyAssertion( a:hasAge
a:Meg "17"^^xsd:integer while the restriction concerned by the syntax
of DataSomeValuesFrom only appears after, in small,  within the
explanation. This is a (trivial) example of the doc, but  for those
which are longer or more complex, this may be quite disturbing or
confusing for some non advised readers, making it more difficult for
them to catch the right illustrative axiom among the other ones.

So if you finally decide to touch the examples, this might perhaps be
rectified as well. What about improving the presentation by simply
changing the order/place of the  axioms and/or the style (for example
another font, in bold or any else way)?

Christine

2008/9/14 Boris Motik <boris.motik@comlab.ox.ac.uk>:
>
> (I redirected this discussion to public-owl-wg, because I feel this is a more appropriate list.)
>
> Hello,
>
> Thanks a lot for this analysis -- it is certainly important to make the examples as consistent as possible.
>
> Before I change the examples, though, I believe we need to decide on the purpose of the English examples. I included them into the
> spec because I felt that many readers could benefit from an intuitive explanation what a particular axiom means. At first, I tried
> not to use the actual OWL elements in the example; thus, I would explain an axiom
>
> SubClassOf( a:Child a:Person )
>
> with the sentence "Children are people". But then, some people complained about such paraphrasing of the axioms: they felt that this
> was imprecise. Instead, they thought we should paraphrase this axiom as "Each instance of a:Child is an instance of a:Person as
> well" -- that is, to use a more modeling-centric view. I updated much of the spec; however, I did not know myself what to do in many
> cases. Thus, it is highly likely that the examples are inconsistent.
>
> Now the question is really what approach to adopt. I still believe that having some kind of English explanation would be very
> useful. I'd like to hear from others about what kind of approach to adopt there -- a more natural-language one or a more OWL-centric
> one.
>
> Thanks again -- I find this analysis really useful.
>
> Regards,
>
>        Boris
>
>> -----Original Message-----
>> From: public-owl-dev-request@w3.org [mailto:public-owl-dev-request@w3.org] On Behalf Of Kaarel
>> Kaljurand
>> Sent: 14 September 2008 20:26
>> To: public-owl-dev@w3.org
>> Subject: English examples in the OWL 2 syntax specification
>>
>>
>> Hi,
>>
>> I extracted all the examples from the OWL 2 Syntax specification (a
>> revision from
>> the end of August) to see how the specification expresses the OWL
>> axioms in English.
>> After sorting the examples by the axioms, many irregularities in the
>> English expressions
>> were revealed. I think most of the irregularities are unintended/unwanted.
>>
>> See the report:
>>
>> http://www.cl.uzh.ch/kalju/ontologies/OWL_spec/owl_spec_examples.html
>>
>> --
>> kaarel
>
>
>
>



-- 
Christine
Received on Tuesday, 16 September 2008 08:08:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 16 September 2008 08:08:14 GMT