Brontekst bekijken van Overleg:Taxonomy/1. Quality/02. Models


{{!|A.M.||
* ...or otherwise unaccessible: yes, also if it does not exist yet. what about design models?

*Roel said in formalisation the semantic is eliminated and substituted by manipulation rules.
(not that this needs to go into the text, but it fits to what you say about mathematical models)

*hey abstract from everything that is not essential: this is a dangerous statement. possibly they abstract
from something that is essential (see models of human behaviour, social systems, where this issue
is more obvious. i do not believe that all essential things are formalizable).

*We should always be aware of what we abstract from and why: yes agree, but that can be difficult.
at some point Roel asked Jelena to specify all the behaviour that she excluded from her case study,
which she excluded, because it is infinite (we exclude the monster coming in and eating the
controller,...)

*Jelena: Does this also fall into what you call 'abstraction'? Assuming that monster will not eat the plant:) is different than not putting the colour of the belt into the model.  

*in formally defined mathematical structure:
i do not understand the section Structure at all. :-/

*the chinese box principle:
in Roels book i saw a distinction between encapsulation and layering. encapsulation is related to the
chinese boxes.
}}

{{!|Stijn H.||
The distinction between the "physical or mathematical" nature of models I believe to be somewhat limited: cannot a (natural language) text be a model? In conceptual modelling (a rather big field in information systems), this difference is often respected (people talk, for example, about "conceptual" (socio-cognitive language) versus "symbolical" (mathematical language) but then again, aren't regular words symbols, too?). I would personally postpone the formal-informal distinction and talk about "physical or symbolical" (meaning physical or language-based), and then perhaps say some more about the type of language (informal, semi-formal, formal) and the type of representation (linear, schematic, or a combination of both).

Within ICIS, we all agree that formal models are what we focus on. However, if you are concerned with the path(s) to good formal models the way I am, transformations of informal models to formal ones are part of the picture, and therefore, so are informal models (either textual or graphical). We can then proceed by clarifying why NL texts or informal (or even semi-formal) doodles don't suffice for our specific purposes.

In my view, the brief discussion of natural language in the "four worlds section" is OK but mentioning natural language comes a bit late, and it is put in too negative a light. Natural language may be inherently open for more than one interpretation, yes, but not all NL use is "sloppy" per se. A well written text can be an excellent model --up to a point. At ICIS, this indeed is not enough, but as you say yourself, in many cases it is. A vast advantage of NL is, of course, that almost everyone has access to it. It is therefore very useful as a basis for low-threshold formalization.

This is just to play the part expected of me, I guess. I have no strong feelings about this, basically I find the definition and examples very workable, though perhaps a bit fragmented when it comes to presentation.
}}

{{!|Hanno Wupper|| For what I call "mathematical model" I try to distinguish between the model - a mathematical object in a mathematical space - and the formula that is used to pin it down - a piece of text. I find it confusing and misleading to all a text a model unless that text itself, not its semantics, mimicks the thing to be modelled. (In that case it would me a physical model, anyway.)

Having said this, maybe by "mathematical model" I mean the same as what you call "conceptual model". Do we have conceptual models that do ''not'' correspond to a mathematical object? The objects denoted by by a blackboard full of ORM definitely are mathematical objects.
}}