ONT Re: Data Models, Ontologies, Logic
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
JA = Jon Awbrey
JF = Jim Farrugia
MW = Matthew West
JF: Just a few brief comments below.
MW: The best definition I know of "data model"
is "the structure and meaning of data".
This is some distance from the ones you
offer below. See below for specific
comments.
JF, citing C.J. Date:
| A data model is an abstract, self-contained, logical definition
| of the objects, operators, and so forth, that together constitute
| the abstract machine with which users interact. The objects allow
| us to model the structure of the data. The operators allow us to
| model its behavior.
|
| C.J. Date,
|'An Introduction to Database Systems', 7th edition,
| Addison Wesley Longmann, 2000, page 14.
MW: This seems to be how data is represented and processed, rather than
anything to do with a data model (aka entity-relationship model).
JF: I would agree that the above definition seems to deal with how
data are represented. It doesn't seem to me, though, that it
deals with how data are processed, if you mean processed by
an actual implementation.
My problems are here:
1. "Object" = geek-speak for "object code module" --
It continues to be one of the worst ideas in
the history of computing to keep using the
word "object" to refer to a passel'o'signs.
2. Abstract machine.
3. Operators.
4. Behavior? Of the data? More likely the DBMS.
All of this goes way beyond a minimal model of the data.
I am pleasantly recalling my consultant mode.
I would like to go back as close as possible
to the basics, which is why I introduced my
honest to goodness concrete example. More
data than you suspect is gathered in what
we used to call "flat ascii files", before
the DBS or the StatPac even enters the scene.
There is some kind of data definition already
in place at this stage of the game. Roughly:
L c D_1 x ... x D_k
L = relation, D_j = j^th domain.
Nuff said.
Time for dinner,
Jon Awbrey
JF: (In the same book, a few lines below the definition quoted above, is:
"An implementation of a given data model is a physical realization on
a real machine of the components of the abstract machine that together
constitute the model.")
JF: I don't mean to suggest that Date's definition is gospel.
But he is a significant author in database-land, and his
way of defining the phrase "data models" is probably at
least worth considering.
JF: Also, I would like to suggest that maybe Matthew's "structure and
meaning of data" really does show up in Date's definition, in the
following ways.
JF: Matthew's definition deals with the structure and meaning of data.
Perhaps one could sensibly claim that the definitions of objects and
operators (from Date's definition) are a representation that allows one
to model the structure of data, maybe something like the way that molded
clay represents and models a chair? (Is it possible to model the structure
of data without representations?)
JF: And perhaps the meaning of the data is gotten at, at least partially,
by the operations that can be defined on that data.
JF: I'm just guessing here, and trying to see if this way of
understanding Date's definition makes sense to Matthew.
JF: For my part, I'd be happy to see Matthew and Jon
more or less agree on a definition that suits them,
or come to feel that they each know what the other
means by "data model."
JF: I'm more interested in following these ideas back and forth
to see what may come of this discussion than in trying to
foist on you an inappropriate definition.
JF: I'll leave comments on the Vianu article for later.
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o