As I wrote this, I should perhaps try to explain it. I agree it is rather
strange when read out of context. (01)
The section from which it is quoted concerns how to define the idea of
importing one CL text into another. Importing was pioneered by the OWL
languages, and is widely used in the semantic web. One OWL ontology can
'include' the text of another by importing it, referring to it by name in the
imports expression. What this means is never exactly defined by the OWL specs,
but roughly it should mean that the effect of importing should be exactly like
copying the importED text into the exact place in the importING text that the
imports expression occurs. In the CL document the necessary formal machinery of
naming a text is worked out in some detail (base on earlier work on 'named
graphs'), and it required the spec to introduce the ideas of a network and of
publishing texts on a network, and the semantics requires that text names be
rigidly defined over the network. Without these ideas, the semantics of text
naming cannot even be stated. (02)
But there is a slight complication. In the RDF/OWL world there is a global
assumption that all Web ontologies are written in the same formal language - in
this case, OWL. So the idea of simply copying one chunk of OWL text into
another OWL text makes sense: the result would indeed be a syntactically
correct OWL text. But for Common logic, things are not so simple. CL allows
multiple dialects, each with a different syntax. Simply copying a text in one
dialect into a text written in a different dialect could produce a chimera
which was not in either dialect, and would not be legal CL. So there has to be
a condition which makes the importing operation make sense. Roughly, it must be
the case that all the texts on a given network are written in the same dialect,
or that some means is provided (for example, an automatic syntactic translation
mapping) which enables any two texts to be included on in the other. Further,
when this is done, the resulting text will be interpreted in such a way that
all quantifiers in it will range over the same universe of discourse. (This is
just a normal consequence of the usual semantic rules for interpreting texts.)
Which means that if one of the texts - say, the imported text - was intended to
be understood as being 'about' a different universe of discourse than the
importing text, then the result of the importation would no longer have the
meaning intended by the author who published it. (This is a real issue on the
Web, in fact: OWL ontologies have no way to indicate that they are supposed to
be understood as being 'about' a limited universe, even though most of them
were in fact written with such a restricted universe in mind. This is the
'horatio problem' : there are more things in heaven and earth than are dreamt
of in your ontology.) Since the whole point here is to write the specification
in such a way that CL texts do NOT change their meaning when transmitted over a
network, this condition is imposed to prohibit it. It is a condition on the
network itself, rather than on the logic; and the only other way to achieve the
desired result would be to prohibit importing itself; which rather takes away
the entire point of having a network in the first place. Also, bear in mind
that CL provides a way to publish content which does not make this assumption,
by making it into a module (which names its own local universe) rather than
simply as a bare text. (03)
I hope this helps. (04)
Pat Hayes (05)
On Jan 23, 2011, at 7:12 PM, Tara Athan wrote: (06)
> As a tangent to the discussion on context and interpretations. I have a
> question about a requirement I just read in the Common Logic Specs
>
>
>http://standards.iso.org/ittf/PubliclyAvailableStandards/c039175_ISO_IEC_24707_2007(E).zip
>
> All texts which are published and identified on a network shall be
> mutually interpretable with all other
> texts on the network which can import them, over the same universe of
> reference and domain of discourse,
> and with their vocabularies merged. This condition applies to all texts
> which might possibly import other texts,
> even if they do not in fact do so in a particular state of the network.
>
> I assume this applies to the Internet as well. This seems to be a
> rather draconian requirement, an opt-out strategy rather than opt-in.
> In a limited setting, such as a sequence of ontologies capturing
> interpretations intended to be snapshots or scenarios all controlled by
> the same publisher, then it would not be so difficult to add mutual
> incompatibility annotations to all these ontologies.
>
> But in a broader setting, there will be incompatibilities across
> publishers that it seems to me will cause difficulties.
>
> Perhaps I have misunderstood- I will be grateful for any clarification.
>
> Tara
>
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/
> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>
> (07)
------------------------------------------------------------
IHMC (850)434 8903 or (650)494 3973
40 South Alcaniz St. (850)202 4416 office
Pensacola (850)202 4440 fax
FL 32502 (850)291 0667 mobile
phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes (08)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (09)
|