[comp.databases] Integration of data formats

tcamp@uncecs.edu (Ted A. Campbell) (03/29/91)

(from TCS BITS -- newsletter of the Triangle Computer Society,
Raleigh--Durham--Chapel Hill, NC)



WHAT'S AVAILABLE
FOR DATA INTEGRATION?

A Query

This month I would like to interrupt my usual tirade of potshots
and clever suggestions to ask a question. How can we integrate
and link data of different sorts? 

Let me give the particular example that prompts this question.
The Space Flight Simulator utilizes a number of different data
formats to describe stars and planets and satellites: *.fd files
contain "focal data", i.e., specific data on a particular
orbital focus (a star, planet, or large natural satellite of a
planet that could be orbited by a spacecraft); *.spd files
contain "spherical projection data", i.e., data in the form of
latitude, longitude and altitude coordinates by which places can
be located on a spherical surface (each orbital focus typically
has a linked *.spd file that describes the surface of the planet
or satellite). 

Two matters might be noted about these data files. In the first
place, they are presently linked together directly, with one
line in the *.fd file specifying the filename for a linked *.spd
file. Secondly, there are other linkages between the data that
could be made: for instance, I could link somehow all the
planets in a stellar system (unfortunately we only have data on
one), and then each planet might have linked data on its
satellites. What I am beginning to see the need for is some way
to link data of differing formats: the *.fd files have a
specific number of fields; the *.spd files have fields for as
many points on (or above or below) the surface as one needs to
describe. 

Moreover, I am thinking, could I somehow link pure (ASCII?) text
to this data? Could there be a text file briefly describing each
orbital focus in addition to the *.fd file for it? Could there be
text associated with specific points in the *.spd file? Then as
the spacecraft entered the gravitational field of an orbital
focus, the program could display text about the focus, and as the
spacecraft passed over specific points on the surface of the
orbital focus, the program could give some details about this
point in text format. 

Of course, the data could be linked simply file-to-file, but I am
wondering if there could be a more universal linkage system. I
have seen hypertext systems that utilize text databases wth
linked items, and some of these can even integrate text and
graphics. I am wondering, though, if there are existing
mechanisms by which wildly different data--text, spreadsheets,
databases of various sorts, graphics of various formats,
etc.--could be linked to each other.

The idea I have been toying with is what I would describe as a
"data socket": a single file that describes another file (or i/o
stream), and whose purpose it is to give in a somewhat more
universal format a description of the data available in the
target file and (more importantly) to link its data to other
data files (or streams) with different data formats relevant to
the data in the target file. Thus, a "data socket" file might
describe a star system (the solar system), but could somehow
explain links to datafiles that describe planetary surfaces,
planetary satellite systems, text files with planetary
information and description, and perhaps graphics files with
images of the planets described. Perhaps, then, the data socket
could also point in another direction to a data socket
describing (data about) the star around which this planetary
system is organized, and this could be linked in turn to data
sockets pointing to information about other stars. 

Could we link data to data and build an, er, uh, "Encyclopedia
Galactica" (apologies to Isaac Asimov)? Could I eventually link
(through a long chain of data sockets) my Space Flight data to
the eighteenth-century religious data that I have in quite a
different area in my computer? The trick, it seems to me, is to
find a near-universal way of describing and then linking data of
wildly different flavors; and I'll await your suggestions and
reflections on this with considerable interest.

(TCS Bits readers: this column is being posted simultaneously to
a couple of newsgroups on the Internet as well.)

- Ted A. Campbell
  Duke Divinity School
  internet: tcamp@uncecs.edu