[tmql-wg] Proposed new requirement: Ability to produce textual output

Chris Angus Chris.Angus@kalido.com
Wed, 2 Jul 2003 10:48:33 +0100

* Chris Angus
| A significant part of the abstract model for output would equate,
| therefore, to that model. =20

* Lars Marius Garshol
| That was an interesting idea. So you mean that the output system would
| essentially be a transformation of result sets into output? From what
| you wrote further below it sounded as if you had an idea about an
| abstract model of the output itself. What do you think that could look
| like?

I am a little loathe to say what it might look like at this stage,
principally because I think that the nature of an abstract model is that
it should often represent the result of a process of abstraction and I
am not sure that we are sufficiently down the road yet to know with more
precision what it is we are abstracting from.  In my experience there is
a danger in choosing the abstraction too early and allowing it to force
restrictions on the solution.

One thing that I wouldn't rule out at this stage, however, is the notion
that the abstract model might not itself be based directly on the TM
conceptual model.  People have talked about the output of TMQL being a
TM, by which I have presumed that they were thinking along the line of
SQL where the output is a relation that represents the net result of
applying relational algebra to the underlying relations.  But suppose
that we introduced some new PSI's to represent the additional constructs
that we wished to have in our output model?  The topics and associations
would then not be restricted to those of the topic maps being searched
but would include the result set itself, lists/sets of topics,
presentation info, etc.  I guess that you can see where I am going with

Chris Angus
Product Architect
Kalido Ltd
Direct:  +44 (0)20 7934 4960
Home:  +44 (0)16 9774 1504
Fax:     +44 (0)20 7934 3377