SUO: Irreducible Triadicity
Dear Colleagues,
This has been a very interesting discussion, in which I have learnt some
things, and been surprised at some things that I thought were common ground
but were not.
Firstly the good news. I think I now understand what John means when he says
"mediating". He means an activity, or the result of an activity, in which
more than two things are necessarily involved form an ontological viewpoint.
Pat has explained some theoretical background to graph theory from which I
draw (actually I already knew this) that there is more than one way that the
world can be represented as a graph (or some other formalism) and that the
triadic iirreducibility only applies when you take one particular
representation.
I have one unanswered question. Part of the claim about irreducible
triadicity is that anything that is higher than triadic can be reduced to
triadic. Pat and Nicola (I think) have suggested that there is no reason why
the limit should be 3 for ontological dependence. So I have presented a test
example about measurement. Measurement is just about the most complicated
thing I have come across in my years of analysis, and I am prepared to
believe that if the information about a measurement can be reduced to triads
that are independent, then it is likely that anything else can. On the other
hand John has not responded to this example, which I repeat below.
> MW: OK. In that case please show me how a measurement can be
> reduced to triads. A Measurement is an activity that:
>
> has a subject (the thing measured)
> an observer (doing the measuring)
> an instrument (e.g. thermometer)
> a measure (what about the thing is being measured, e.g. temperature)
> a result (the value e.g. 20)
> a Unit of Measure (e.g. Celsius)
> the time when the measurement was taken.
I would be interested to see whether and how this can be reduced to triadic
or less ontologically dependent elements. If it can't, then there is at
least one case where this does not apply (which is enough) or perhaps as Pat
has been suggesting there has been confusion between graph theory and
ontology.
Well I guess the next question is "does it really matter?". By this I mean,
does it really matter that there are different mappings from the world onto
graph theoretic representations.
Unfortunately the answer is "yes", as we discovered in Shell some 10-15
years ago. At about that time many of the 200 or so companies in the Shell
Group were getting into relational databases, mostly developing much the
same systems to do much the same thing in different countries. Something
that we noticed to our horror was that all these systems were generally
incompatible, and that generally they were not transferrable either. We
(then the corporate data management group) were asked to look at why this
was the case.
The databases were generally documented in data models. (Data Models can be
seen as graphs with entity types as nodes, relationship types as edges, and
attribute types as leaf nodes.) When we examined the data models we found
that one of the (many) problems was that people had mapped the real world
onto these constructs in a variety of ways. Let us take activities as an
example:
1) as a relationship type, with the participants at either end of the
relationship.
2) as a "transaction" - an entity type with a number of dependent
relationships.
3) as an independent entity type with non-dependent relationships to the
objects involved.
4) a mixture of 2 & 3.
1)
o------------o o----------o
| | throws | |
| person |>-------------<| ball |
| | | |
o------------o o----------o
2)
o------------o o----------o
| | thrower | |
| |>--------------| person |
| throws | | |
| | o----------o
| | o----------o
| | thrown | |
| |>--------------| ball |
| | | |
o------------o o----------o
and even 3)
o------------o o----------o o----------o
| | | | | |
| |--------------<| thrower |>---------| person |
| throws | | | | |
| | o----------o o----------o
| | o----------o o----------o
| | | | | |
| |--------------<| thrown |>---------| ball |
| | | | | |
o------------o o----------o o----------o
Of course these different structures are not directly compatible
and in many cases arbitrary combinations of these different approaches
were taken.
So one of the things we did was to study the differences between these
(and other) representations. The conclusions to these have been
implemented in the EPISTLE Core Model. Incidently, the ECM V4.0
has just been published, and can be found at:
http://www.stepcom.ncl.ac.uk/epistle/data/epistle_core_model_version_4.htm
Regards
Matthew
===============================================================
Matthew West http://www.matthew-west.org.uk/
Principal Consultant Shell Visiting Professor
Operations & Asset Management The Keyworth Institute
Shell Services International The University of Leeds
http://www.shellservices.com/ http://www.keyworth.leeds.ac.uk/
H3229, Shell Centre, London, SE1 7NA, UK.
Tel: +44 207 934 4490 Fax: 7929 Mobile: +44 7796 336538
===============================================================