Lists Home |
Date Index |
A priori information reduces uncertainty (basic concept)
for any Markov process (how are probabilities for initial
states and transitions chosen is an interesting subject).
A schema is a priori information correlated on receipt of
information (message validation). It acts as a filter
and a control to reduce noise by enabling feedback. A
schema can be thought of as a signature or intelligence
based on memory or forethought. If we could overload
a schema the way we do a function, it might be interesting
and a step forward in the "not THE schema" discussion.
The theoretical scalability of the WWW is predicated on blind
exchanges. (The protocol may be stateless but that doesn't
mean a SYSTEM is, so we use messages to convey state. That's REST.)
The practical reality is that few exchanges are blind and
that WWW scalability is seldom required. Web services
relax the protocol constraint but tighten the requirement
for schema-aware exchange to enhance the chances of success
for long running transactions (puts virtual state of
transitions back at the center of the design). In this,
schemas are to messages what the probability matrices
are to Markov models.
Unequal probability is exactly the point. Engineers
do all that they can to make it as unequal as they
can. Never give a RESTafarian an even break. :-)
From: Ed Lai [mailto:firstname.lastname@example.org]
The discussion on unequal probability is unnecessary
and only side-tracks the discusssion.
Lots of possible choice is not information, knowing
which particular choice is taken is information.
It is like the lotto ticket numbers, knowing that the
number you picked is just one of the many millions of
possibility is not much of information. Kowning the
actual winning number is information.
The more the choices, the more information you get
when you know which choices.
Without a schema, when you got the actual XML data,
you know which choices it is out of infinity, that is
a lot of information.
With a schema that only allows 4 valid XML instance,
getting the XML data give you the knowledge of which
of the four, that is not a lot of information.
So a schema does reduces the information a XML message
However that comes from knowing the schema, if you
know the schema, then there is less information in the
message, but this comes about because you already
knows a lot, you know the schema. The schema carries a
lot of information, so once you know the schema there
not much more you can know.
Back to the lotto example, if you know the first 5
balls, you already have a lot of information, the
lotto drawing would carry very little information.
Knowing the schema is like knowing the first 5 balls.
So if someone knows the first 5 numbers of the next
lotto winning ticket, please tell me.