[
Lists Home |
Date Index |
Thread Index
]
1. If I am exposing a limited set of queries, I control the
mapping so 1 isn't hard. In other words, it is unlikely I
would expose an entire database to ad hoc querying. People
who ask for that don't understand the can of worms they are
opening.
2. OWL is relatively academic. In other words, in a commercial
setting where I rely on a development framework, I am unlikely
to build core technology if I can get the job done with the
framework. For this case, OWL isn't buying me anything that I
can't do with technology that I already have. This isn't to
put down OWL but to note where it is in the adoption lifecycle
vis a vis commercial development frameworks from major players.
If we were in the data aggregation business in which one has
to construct a warehouse from multiple data base sources that
one doesn't control, the case for schema mapping is better
but unless the numbers of these are high, I am usually better
off enabling a human analyst/developer to do that work based
on human negotiation and an interchange schema, eg, GJXDM.
len
From: Thomas B. Passin [mailto:tpassin@comcast.net]
Bullard, Claude L (Len) wrote:
> It is completely possible to expose the schema or to expose
> an interface schema that middleware then transforms into
> the local schema. As you say, Thomas, mappings. BTW, why would I
> need OWL to implement what is essentially a business object?
If you are interested in describing your schema - which could well be
the schema of special views or reports that you provide - you need to
use some language. Mappings can be constructed when two kinds of
information are known -
1) Identity - which primary key of one entity is equivalent to which key
of another;
2) The semantics of terms and their relationships.
1) is hard, and has no general solution, but only specific ones. For 2)
why not use some standard language, and if you do that, why not use OWL?
OWL would not be "implementing" anything, but would contain needed
data for the actual implementer.
|
