ONT Re: Data Models, Ontologies, Logic
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
Note 15
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
JA = Jon Awbrey
JF = Jim Farrugia
JF: Two questions below, about the shape of the data.
JA: "So what does your data look like?"
JA: "What is the meaning of this code?"
JF: Are we assuming that the user comes in with the file header
and the columns of data? If so, what does the user want to
do with the data, and how does modeling the data as records
in the space X help her achieve her goal?
It's not supposed to, but it very often happens that way.
An international health practitioner/researcher just got
in from a 2-year stint in the Sudan, with all her data
in a 5-yr old spreadsheet type format. There's a new
"stat person" that's she's never talked to before and
maybe didn't talk to the one that was there before
she left.
The fragment of the FIA file that I posted in my initial message
is pretty much verbatim what I was handed in 1989 for my own work
on "Exploratory Analysis of Sequential Interactions" (EASI).
The FIA data that we are looking at came out of major funded research (NIMH),
a horde of researchers from a particular cross-disciplinary mix of paradigms,
a long history of developing constructs, a multitude of research hypotheses,
and a host of different objectives. Think of Census Data. What's it for?
This was slightly more structured than that, with some teams using control
groups and experimental groups, and others testing specific hypotheses,
but it was pretty much an exercise in omnibus data gathering overall.
I was lent a sample of the data to explore a few "radical" ideas of my own
involving adaptive formal language learning and analytic logical modelling.
I only got a hint of the actual complexity of the mother-ship, since it is
not unusual to shield low-level data miners from the actual hypotheses and
since I was doing exploratory data analysis, I did not really care to know.
But the generic question would be "What is going on here?"
That is, can you detect salient patterns in the data, some
of which may tell you something about the way that people
behave in general, and some of which may help you to fix
a panoply of social problems, like the effects of alcohol
dependency in parents on the developmental health of the
children, or any number of things like that down the road.
In sum, the space X is only a "data base", quite literally,
a base of data from which you may think to compute and to
derive any other statistics that you think will help you
to answer the basic questions: "What is going on here?"
& "What, if anything, could, or should, we do about it?"
Jon Awbrey
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o