W3C home > Mailing lists > Public > public-semweb-lifesci@w3.org > May 2007

Re: Advancing translational research with the Semantic Web

From: Pat Hayes <phayes@ihmc.us>
Date: Thu, 17 May 2007 09:57:31 -0700
Message-Id: <p0623092dc27235a932c0@[192.168.1.4]>
To: samwald@gmx.at
Cc: public-semweb-lifesci@w3.org

>  > >Statements about believe, evidence and provenance can be easily
>>  >attached to "<binding_process>".
>>
>>  Hmm. They could, but that isn't really internally coherent. If this
>>  is supposed to be the name of a process, then its doesn't make sense
>>  to say that it has a provenance or that it is subject to belief.
>
>That depends on the semantics of the relations 
>you use to relate the biological entity with 
>believe, evidence and provenance. Change "this 
>statement is believed by..." to "the existence 
>of this entity is believed by...", change "this 
>statement was derived from document..." to "the 
>existence of this entity is also described by 
>document..." and so on.

That will get you so far, but you have to be VERY 
careful. Will other uses of these relations 
retain these very refined meanings this 
precisely? It is notoriously easy to get muddled 
between use and mention issues.

It doesn't really work, in any case. Suppose 
someone wants to say that an assertion made by C 
is wrong because it mis-uses a name to refer to 
the wrong entity. There is no way to even express 
that kind of reason for deprecating an old 
assertion, using this trick, since you have made 
the meta-data refer to the entity rather than to 
the name. And in any case, your transcriptions do 
not capture the right meaning. To say that 
document A describes the same entities as 
document B is not to say that A was derived 
*from* B. For one thing, 
describing-the-same-entities-as is symmetric.

Moreover, the appropriate logic of these belief 
relations is probably not the same as that of 
relations which are free of propositional intent. 
This will not matter for pure RDF use, since RDF 
is so logically weak, but it will as soon as you 
start using it in conjunction with OWL.

>  > >We have already done this for some ontologies we developed for the
>  > >Banff demo. I think that this approach will proof to be sufficient
>>  >for most use cases,
>>
>>  It will work for a while, then it will break when things get more
>>  complicated. Its a hack; hacks work, but one shouldn't forget that
>>  they are hacks,
>
>This is not a hack at all when you use the properties I described above.

Im afraid it is, in my view. One cannot get a 
satisfactory logic of propositional attitudes 
(such as belief) by burying them into 'opaque 
relations' and using a conventional assertional 
logic. This has been tried, and is known to not 
work.

>It is fully inside standard RDF and OWL semantics

RDF, yes: OWL no. The problem arises from 
equality substitution, and owl:sameAs is equality.

>and can be expressed in RDF/XML without any 
>problems. Introducing thousands of tiny RDF 
>files just to be able to refer to the statements 
>separately -- THAT is a hack and will break when 
>things get more complicated.

Brevity does not make something un-hackish. Often 
the short snappy way to do something turns out to 
be the way that breaks.

>The approach I described above should scale without problems.

I predict that it will not scale to OWL, unless 
you are very careful to exclude the parts of OWL 
that allow identity conditions to be stated. In 
fact I also predict that as soon as you have more 
than, say, 5 users composing content, they will 
start to disagree about how to use these 
'psychological' relations appropriately.

---------

But OK, let me not be too picky. I agree that 
this might well work for a while, and its better 
than nothing, and it gets the show on the road. I 
didn't mean to start a huge argument. Juast make 
sure that when y'all find a better way to do it, 
you don't re-use these relations with a new 
meaning. Use a brand-new namespace instead.

Pat

>
>cheers,
>Matthias Samwald
>
>----------
>
>Yale Center for Medical Informatics, New Haven /
>Section on Medical Expert and Knowledge-Based Systems, Vienna /
>http://neuroscientific.net
>--
>Psssst! Schon vom neuen GMX MultiMessenger gehört?
>Der kanns mit allen: http://www.gmx.net/de/go/multimessenger


-- 
---------------------------------------------------------------------
IHMC		(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32502			(850)291 0667    cell
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Thursday, 17 May 2007 16:57:47 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:20:26 UTC