RE: Upper ontologies

Please allow me, as a co-author, to tell you some about one of the 17 upper
ontologies on the Wikipedia list: ISO 15926-2.

 

Its domain is the world of the process industries, with all parties involved
in them during the lifetime of a facility. Together with its other parts it
allows for representing the life-cycle information about a facility, its
components, its streams and its activities, both process activities and
human activities. Storing life-cycle information calls for a 4D approach
with the concept of temporal parts. This information covers facts, and has
to originate in the applications used during the life of the facility. These
apps can launch SPARQL queries to fetch input data, as earlier uploaded by
other apps.

 

This upper ontology <https://15926.org/topics/data-model/index.htm>  has 201
generic entity types of which 106 classes and 48 reified relationships and
47 classes of relationship.

 

That model is extended by, what we call, a reference data library
<http://data.15926.org/rdl/RDS416834>  with some 22,000 generic,
domain-relevant, concept classes such as pump, pipe, valve, instrument,
document, status, etc.

Then we have various extensions with classes of standardization bodies (e.g.
ASME, DIN) and suppliers, and finally the plant life-cycle information with
individuals.

Coupled with a different extension, covering concepts of another domain
(e.g. nuclear or pulp & paper or healthcare), the standard can be used for
that domain.

 

All classes are in one taxonomy with Thing at the top.

 

Information is represented with 'templates
<http://15926.org/home/15926_template_specs.php> ', small ontologies using
UO concepts. When designing their concept we were inspired by the W3C
Working Group Note 'Defining N-ary Relations on the Semantic Web
<https://www.w3.org/TR/swbp-n-aryRelations/> ' by Natasha Noy and Alan
Rector.

 

I recently mapped the data model and reference data of CFIHOS, the data
project of IOGP (International Association of Oil & Gas Producers), and
didn't encounter mapping problems. So it is safe to say that it is 'fit for
purpose'.

 

As a noteworthy feature we adopted the Possible Worlds
<https://15926.org/topics/possible-worlds/index.htm>  of David K. Lewis
<https://en.wikipedia.org/wiki/David_Lewis_(philosopher)> ' treatise "On the
Plurality of Worlds <https://plato.stanford.edu/entries/david-lewis/> ",
allowing us to model engineered objects and real world objects in an
identical manner, yet separated in their respective worlds. A perfect
starting point for Digital Twins.

 

If interested, visit 15926.org <https://15926.org/home/>  

 

Hans 

____________________________________________________

-----Original Message-----

From: Mark Wallace <mark.wallace@semanticarts.com> 

Sent: dinsdag 12 januari 2021 18:50

To: janowicz@ucsb.edu; Neil McNaughton <neilmcn@oilit.com>; Martin Thomas
Horsch <horsch@inprodat.de>; semantic-web@w3.org

Subject: RE: Upper ontologies

 

Krzysztof Janowicz wrote:

>  meaning is in our heads, not in the world.

 

This in itself is a debated philosophical position. :D

 

-----Original Message-----

From: Krzysztof Janowicz <janowicz@ucsb.edu> 

Sent: Tuesday, January 12, 2021 11:52 AM

To: Neil McNaughton <neilmcn@oilit.com>; Martin Thomas Horsch
<horsch@inprodat.de>; semantic-web@w3.org

Subject: Re: Upper ontologies

 

 

On 1/12/21 2:22 AM, Neil McNaughton wrote:

> Does this not imply that every time you move up a level in an ontology
there are "different points of view have advantages and disadvantages". If
an upper level ontology is impossible, why are levels top-1, top-2 etc.
doable?

 

 

Science and society are in constant flux. Meaning is an emerging and
evolving property of (human) cognition both on the individual and societal
level. Put differently, meaning is in our heads, not in the world. Meaning
cannot be fixed in ontologies. Instead, ontologies aim at restricting the
interpretation of domain terminology towards their intended meaning. Such
restriction aims to maximize semantic interoperability, i.e., to minimize
cases where parties exchange information that seems valid on a syntactic
level but where the expected semantics of the target does not match the
semantics of the source.

 

Once we leave the ground of concrete domains and their tasks, we lose the
context that helps restrict a domain's terminology and a sense for the
required quality of this restriction, e.g., the choice of axioms. As a
result, top-level ontology becomes metaphysics - useful, but not easily
ground-able. One top-level ontology may declare that there are two disjoint
kinds: objects and events. A second such ontology may declare that 'objects
are just very slowly evolving events'. You can select the one you prefer,
but none of them is wrong. It is also often unclear when and why one would
have to restrict the interpretation of terminology for a certain, concrete
task, e.g., data retrieval, by going all the way up the abstraction
hierarchy.

 

Personally, I often search for measurement types and face the semantic
interoperability challenge that two different measurement procedures go by
the same name. I rarely search for things that are 'substances',
'occurrences', or 'forms'. All that said, I fully understand the
intellectual joy of designing and studying these top-level ontologies, but I
see their use in providing us with answers to the big questions of what can
be said and which distinctions make sense and less on the level of
actionable theories.

 

Jano

 

 

> 

> Best regards,

> Neil McNaughton

> Editor Oil IT Journal - www.oilit.com

> The Data Room SAS

> 7 Rue des Verrieres

> 92310 Sevres, France

> Landline+33146239596

> Cell+33672712642

> 

> 

> 

> -----Original Message-----

> From: Martin Thomas Horsch <horsch@inprodat.de>

> Sent: Monday, 11 January 2021 17:04

> To: semantic-web@w3.org

> Subject: Re: Upper ontologies

> 

> Dear Mikael,

> 

> four reasons. The first one applies to all standardization, and to all
metadata standards, not only top-level ontologies, and it is simply that
multiple concurrent development efforts exist. Here, as in many cases, there
is no authority that can enforce the uptake of a single standard.

> 

> Second: By selecting a top-level ontology, you commit yourself to a
philosophical position on what sort of things can exist in the world.

> Unsurprisingly, it is impossible - and undesirable - to make everybody
accept the same point of view. Also, different points of view have
advantages and disadvantages depending on what exactly you want to do.

> 

> Third: It is also not a major problem. At the level of domains of
knowledge, there will always be semantic heterogeneity, or in other words,
multiple domain ontologies will be in use at least in some domains.
Therefore, solutions addressing this heterogeneity need to be co-developed
with any major innovation in data management anyway.

> 

> Fourth: Once a top-level ontology has been developed, it can never
disappear. Even if most of the community recommended one of them, the others
would still be around, and people would be able to use them.

> 

> Best wishes,

> Martin

> 

> On 11/01/2021 16:40, Mikael Pesonen wrote:

>> Maybe this is a stupid question but why is there (at the moment) 17 

>> different upper ontologies:

>> 

>> https://en.wikipedia.org/wiki/Upper_ontology

>> 

>> Isn't the idea to make just one that everyone can use?

>> 

>> 

>> 

> 

 

--

Krzysztof Janowicz

 

Geography Department, University of California, Santa Barbara

4830 Ellison Hall, Santa Barbara, CA 93106-4060

 

Email: jano@geog.ucsb.edu

Webpage: http://geog.ucsb.edu/~jano/

Semantic Web Journal: http://www.semantic-web-journal.net