RE: SUO: Why is IFF and the Lattice of Theories good for SUMO
Dear Adam,
See reponses below.
Matthew West
Principal Consultant
Shell Information Technology International Limited
Shell Centre, London SE1 7NA, United Kingdom
Tel: +44 20 7934 4490 Other Tel: +44 7796 336538
Email: matthew.west@shell.com
Internet: http://www.shell.com
> -----Original Message-----
> From: Adam Pease [mailto:apease@ks.teknowledge.com]
> Sent: 07 May 2003 18:14
> To: West, Matthew R SITI-ITPSIE; Standard-Upper-Ontology (E-mail)
> Subject: Re: SUO: Why is IFF and the Lattice of Theories good for SUMO
>
>
> Matthew,
> I agree broadly with these points, but the questions that
> remain are
>
> 1. What tools or methods can be demonstrated on a problem in
> mapping, and
> what are the advantages of one approach (as shown in an
> example), over another
MW: In ordinary database systems mapping has been going on for more than
a decade. The maps between models are more usually known as interfaces,
and in Shell at least a survey showed in the mid 90's that they accounted
for 25-75% of the cost of new systems. Today there are systems that are
dedicated to managing iterfaces, they often go under the name of ETL
(Extract Translate Load) or EIA (Enterprise Integration Application).
Where the former is for batch/transaction data and the latter for online
federation of systems.
>
> 2. Where are the particular concrete proposals of different
> theories? It's fine to mention that such may exist, and certainly
> different theories have been described in the literature, but until a
> particular one is proposed that replaces a defined portion of
> another, this
> seems rather speculative for me. One could define an
> alternative module
> for SUMO, but no one has done so. Has something of that sort
> been done in
> EPISTLE?
MW: We have certainly made changes to the way that EPISTLE models change.
http://www.matthew-west.org.uk/Documents/InformationModellingPDT2002.pdf
This brief paper gives an overview of how the two approaches relate to
each other. Many implementations are still using the old approach, and
for some purposes the old way may be better.
>
> Adam
>
> At 01:17 PM 5/2/2003 +0100, West, Matthew R SITI-ITPSIE wrote:
>
> >Dear Adam,
> >
> >I could of course have titled this note "Why is IFF and the
> Lattice of
> >Theories good for EPISTLE" since we are in a similar position, having
> >a single coherent ontology, though focussed on a different paradigm
> >and problem set than SUMO as far as I can see.
> >
> >There are two pragmatic reasons why I support this approach:
> >
> >1. Version Management
> >
> >In the great scheme of things all our present day ontologies are
> >immature, which means that they will change and improve over time.
> >Even if they were mature, knowledge is expanding at an alarming rate.
> >
> >This means that we can expect our ontologies to change. The problem
> >with this is managing the change in the ontologies so that you can
> >track what version you used for some problem, and how the current
> >version is different from this.
> >
> >The lattice of theories provides facilities for this. You
> can identify
> >what theories were parts of what versions of the ontology.
> >
> >
> >2. Using multiple ontologies/paradigms
> >
> >My personal view is that the "holy wars" about different sorts of
> >ontology (e.g. 3D/4D) are largely misplaced. I expect that it will
> >turn out that different ontologies/paradigms will have different
> >strengths/weaknesses and that we will want to translate between
> >these for more complex problems.
> >
> >IFF and the Lattice of Theories supports this. The support here is
> >weaker, because in many cases you would need to do the mapping
> >between at least core constructs of equivalent theories, but an
> >environment that supports this is still valuable.
> >
> >
> >Matthew West
> >Principal Consultant
> >Shell Information Technology International Limited
> >Shell Centre, London SE1 7NA, United Kingdom
> >
> >Tel: +44 20 7934 4490 Other Tel: +44 7796 336538
> >Email: matthew.west@shell.com
> >Internet: http://www.shell.com
>
>