W3C home > Mailing lists > Public > public-prov-wg@w3.org > July 2011

Models and their use

From: Simon Miles <simon.miles@kcl.ac.uk>
Date: Tue, 12 Jul 2011 16:51:01 +0100
Message-ID: <CAKc1nHdzP81AH6PA16Ck3wC1ArCaNQtuyuA6ei=R_ZKDcAf1_g@mail.gmail.com>
To: Provenance Working Group WG <public-prov-wg@w3.org>

I raised a point with Luc and Jim (McC) at the F2F1, but did not have
time to mention it in discussion. It primarily concerns the definition
of Thing/Stuff/Entity/Entity State/BOB.

I think complexity and confusion arises from talking about "modelling"
and "representing" in the model itself. The concepts comprising a
model should be just whatever is modelled, e.g. "process execution",
not how the model is used, e.g. "representation of a process

It should already be clearly understood by users, but can be stated
explicitly, that in using any model you are asserting something about
whatever is modelled using some representation of it. We may also add
that any assertion is by necessity from some perspective, of which
there may be many, and not necessarily objectively true. These points
are separate from any concept definition.

A connected point is that definitions in models will use colloquial
synonyms to get across what a concept is, and these do not need to be
defined, e.g. we do not define "activity" even though we say "a
process execution is an activity..."

I think we got the above wrong for Thing, in saying:
 "things represent real-world stuffs and have properties modeling
aspects of stuff states"
changed to:
 "BOBs represent real-world entities and have properties modeling
aspects of entity states"

First, this definition uses terms "represent" and "modeling", implying
it is about the use of the model not what is modelled. Second, the F2F
discussion ended up with "entity", the replacement of "stuff", being
treated as a concept in the model itself rather than a colloquial
synonym used for the purpose of definition, e.g. we discussed whether
an agent is an entity or an entity state (or a BOB).

I argue we should be clear that there is only one concept being
defined here not two, and it is something in the world not a
representation of it. If we need to, we can say how users should apply
the model in accompanying notes.

A similar problem may affect the IPV-of definition, where properties
are referred to having "corresponding" values. I think they have the
*same* values, which may be represented in different ways. The former
is a constraint of the model, while the latter is a given truth about
the use of any model.

To understand the consequences of the above points, I suggest
alternative definitions at the link below:


Dr Simon Miles
Lecturer, Department of Informatics
Kings College London, WC2R 2LS, UK
+44 (0)20 7848 1166
Received on Tuesday, 12 July 2011 15:51:28 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:50:57 UTC