[topicmapmail] Re: Topic map design guidelines? (Jan Algermissen)

[Lars Marius Garshol]

* Carlo Moneti
|
| I tried modeling family information as well a while back, and found
| it cumbersome to model all of the possible associations. The problem
| reminded me vaguely of an un-normalized relational database.

It is similar! The problem with all the extra association types is
that they are redundant, and that's why I used inference rules
instead. This lets you traverse the relationships as though they were
associations, but without actually having the associations in there.

| Anyway, it occurred to me that only a modest number of associations
| were necessary for "me" to infer the other possible associations.
| When accessing a normalized database, the "database programmer" does
| the inferencing--the result of which is embodied in the query--that
| can produce a sophisticated and informative report.

Bingo. :-)
 
| I envision a day when a generic topic map viewer will be the main
| application we use (kind of like web browsers today); each topic map
| may come with a specifically defined skin (layout template) to
| properly present that topic map's information; functional plugins
| may be called on to manipulate/add/delete data. Perhaps a path to
| total application integration?

Actually, what you are describing here is not too different from what
quite a few of us would like to see TMCL be able to do. The first
draft probably won't have this kind of thing in it, but it may appear
later.
 
| Anyway, an inference engine would be part of the envisoned generic
| topic map viewer. This suggests a need for standards that define
| what the inference engine does and how to interact with it.  Has
| there been any work on such an engine or standards for it?

Oh, yes. :-)

The tolog query language lets you define your own inference rules, and
this has been implemented both in the OKS and in TM4J. Several
end-user applications based on the OKS make heavy use of inference
rules to do their thing. One of them uses inference rules to hide the
versioning that is represented in the topic map, and then builds on
those inference rules to define complex behaviours.

At present it looks like this kind of thing will make it into TMQL[1],
and probably TMCL will make use of some of these facilities, but
exactly which ones and how is not yet decided.

[1] <URL: http://www.jtc1sc34.org/repository/0502.pdf >

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
GSM: +47 98 21 55 50                  <URL: http://www.garshol.priv.no >