[topicmapmail] webservice notification, ws-topics 1.0

Robert Barta rho@bigpond.net.au
Mon, 27 Sep 2004 07:59:11 +1000


On Tue, Jun 01, 2004 at 06:11:05PM +0200, Stefan Lischke wrote:
> btw. is there any research about tmql containment problem (if the result
> of tmql query A is a superset of the result of another tmql query B,
> then query A contains query B)

Sorry 4 the VERY late response.

Stefan,

Yes, there are some bits and pieces answering the subsumption problem
for TMQL.

In short, a TMQL statement can be thought of having an "incoming" and
an "outgoing" side.  On the incoming side you use predicates, path
expressions, etc. to detect what you are interested in a particular
map (or a set of maps). On the outgoing side, you would construct the
result document, which - in the case of TMQL - will be quite likely to
be either a list (ala SQL), an XML instance (ala XQuery) or a map.

If you skim over the formalism

   [ Warning: WORK IN PROGRESS ]

   http://astma.it.bond.edu.au/junk/tau-model.pdf

(section 3), then it is quite simple to define a relationship <=
between two queries q_1, q_2 and a given map:

   q_1 <= q_2    <=>    m * q_1 <= m * q_2

This definition _ONLY_ respects the incoming side using lists of
tuples. We have not yet thought about including XML or map content.

\rho