[
Lists Home |
Date Index |
Thread Index
]
Yes, and that is why in HumanML, I asked for it to be
quickly restricted to human-to-human communications.
It keeps it from becoming CYC or just RDF. You are
working on MachineML: the categories of polite
machine intercourse.
I need to know what you mean by interoperability. As I've
said repeatedly, data is portable, systems interoperate.
My perspectives are thus:
Bits on the wire are portable given equally functional
parsers on both ends. The simplest approach to this
so far that scales structurally, enables names, and
then avoids collisions is XML + Namespaces. Then XML
turns a blind eye to the semantics of the names and
the values of data inside the structures leaving that
to the object model of the type handler. If systems
(implementations of these models) must interoperate,
they need portable data (bits on the wire). But we
still don't have a definition of 'interoperation',
just 'operation'. What does it mean to 'inter'operate?
I suspect we are back to ontological commitment: given
a message (bits on the wire) of some type (in accordance
with an authoritative specification), the observable
receiver will emit a recognizable behavior to which
the observing system may emit a recognizable response.
Mutuality requires perspective. Schemas are
a kind of perspective. By introducing them
to the process, one has introduced the categories
and then verifies the commitments by observation.
What you are assembling is the kinds of information
that are a metric of depth to the commitment.
The trick of the mating game at every age is to be
"willing and eager" given some partner. All of the
noise is in the selection and retention of partners.
To maximize the game, both sides must surrender to
pleasing the other without expectation but with a
knowledge of range of expectation. That is the only way
it becomes "frictionless". This usually does not
begin in humans until they reverse roles at least
once. One alternates being willing in order
to make the other eager.
len
From: Roger L. Costello [mailto:costello@mitre.org]
"Bullard, Claude L (Len)" wrote:
> the attempt
> to create a set of categories of those things/types
> that can affect a human communication.
I like this description, although my interest is in machine-to-machine
(M2M) communication. Thus, I recast it as:
The attempt to create a set of categories of those
things/types that can affect data interoperability.
Comments?
|
