[
Lists Home |
Date Index |
Thread Index
]
If we are to have killer apps, maybe apps that make a
technology grow should be offered as well. What would
give me something useful today:
1. A design group posts design notes semi-annually.
There are say, 85 of these every four to six months.
Proposals, Implementation, Certification and Training
all receive email notifications when these are posted.
But everyone is busy and it can be some time until
they read these. Even if they do, the details down
to the control and field are more than anyone can
master with a single or even repeated readings. They
use these documents for diverse tasks from answering
questions in a proposal, questions in a demo, questions
in a classroom, and so on. Each of those questions
have different levels of formality and impact on the
business process, customer expectations and requirements
and so on. They need agents that read for them and
when asked a question (not a SQL query; a reasonable
representation of a natural language question), returns
a list of URIs that are applicable. In short, if
possible, they'd like an agent that reads the RFP, etc.
and makes recommendations about online company
resources or others that will be pertinent to the
document at hand. (yes, I know about Oracle Context.
It COSTS TOO MUCH, Ellison.)
2. Two programmers at opposite sides of the planet
working a shared project are having a battle of halfwits
to determine whose design is better. Meanwhile the
clock is running and schedules are slipping while
others wait on these MastersOfTheAbstractAndNotVeryImportant
details. Upstream of them, proposal personnel,
sales personnel and contract lawyers have already predetermined
that these designs will interoperate in accordance
with a predetermined schema. An agent that informs
them of this fact and that by the way, the Record
of Authority (ROA) is named THIS URI hosted at a site
where the owners are sharp enough to realize that
a formal URI MUST be assigned to a public record
of authority for a shared process. Don't talk to
me about versioning. Just send the bloody URI.
It is a lot of fun to argue about the abstractions
of resources vs documents, the infinitely recursive
imprecision of describing all of these, the lack of
our control over this random and even hostile universe,
but at the end of the day all the user needs
is a URL to *a file*. If the semweb can't help the
humans find and retrieve the right file at the right
time for the right requirement, it's just more
CaucasoidCaca and we don't need it. If it can,
then bring it on and let's get on with growing
our businesses.
Len Bullard
Senior Technical Consultant
Intergraph Public Safety
|