[
Lists Home |
Date Index |
Thread Index
]
A central issue of any querying system is to define or discover
the appropriate vocabulary for that domain. Which constraint
system forms the basis for the query? For example, querying
a video for frame identifiers is straightforward. Querying a
video for the semantic content requires a classification
vocabulary that can discriminate continuous motion into
sequences of recognizable statements. Then depending on the
formality of the vocabularly, to make associations to other
vocabularies and infer the actual message. Papers on
subjects such as human motion recognition exist.
http://citeseer.nj.nec.com/campbell95recognition.html
It is straightforward to create a vocabulary for any
formal system where formal is that at least one notation exists
and that for any term provided, most experts would
agree that an expression of that term represents the
same meaning (again, the identity problem). I leave
the issues of working with informal or emergent vocabularies
to another time although these are probably not intractable.
We've had to look into this in HumanML (eg, semiotic systems based on
an abstraction for signals, symbols, and signs that enable
a cultural set to be recognized). While I have no new
science to offer, I believe that this is where the SemWeb
and XML have a lot to offer, but that the work required
without good abstraction tools will take a very long time
and be practically undending given semantic drift and
evolution.
Biometric data is a lot simpler and that is why it is
used for final identification. Cultural vocabularies
play a large role in making predictions, creating associations,
etc., but the biometric data is the most reliable for
individual identification.
len
-----Original Message-----
From: Ken North [mailto:ken_north@compuserve.com]
Q. XQuery is document-centric. What can we expect?
How will we query or find information in documents over the next decade?
Not just text -- other rich types including audio, video and biometric data.
|
