[topicmapmail] Triangulation of associations

[Nathan Young -X (natyoung - Artizen at Cisco)]

> As a newcomer to the list and a TM-beginner, I ask humbly for=20
> a piece of
> advice. It may be trivial, but it may also touch on some very=20
> simple form of
> inferencing.

I think some kind of topic map query language might get you what you =
want.

Tolog, tmpath and toma are each interesting in their way and have =
documentation that's sufficiently developed that you can tell a lot of =
thinking has gone on.  I don't know about the availablility of working =
implementations.

http://www.ontopia.net/topicmaps/materials/tolog-spec.html
http://homepage.mac.com/dmitryv/TopicMaps/TMPath/TMPathRevisited.html
http://www.spaceapplications.com/toma/Toma.html

------------>Nathan


> The case is this:
> having made a topic map over my library, let's keep it simple=20
> and focus on
> the three main topic types: person, book, and subject (what=20
> the book is
> about).=20
> There are the obvious association types between them: person=20
> (in the role of
> author) associated with book; book associated with subject, and book
> associated with person (in the role of subject).
> On the instance level, of course, the persons have names, the=20
> books titles,
> and the subjects terms (that are related taxonomically=20
> related, but that is
> an other story).
> The TM is thus populated, everything works well, but one=20
> association is
> annoyingly absent: The one that would give a profile of the=20
> person in terms
> of the subjects associated to the books that person has=20
> authored. In the
> opposite direction, it would also provide the chance to see=20
> the names of the
> who have written on a given topic, as opposed to only the=20
> books on that
> topic.=20
>=20
> Clearly, I couold make a new association type connecting=20
> person and subject
> directly and then type in the appropriate person IDs and=20
> subject terms (or
> IDs), but that would be redundant (and it would involve a lot=20
> of typing).
> The information is already there in the TM. So how do I get=20
> it out in the
> form I want?
>=20
> I came up with the following suggestions (on a conceptual level):
> 1) reify the book-subject association, then make an=20
> association between this
> (new) topic type and person. However, I can't quite figure=20
> out how this
> would get me the right information.
> 2) take the subject role type (which can be played by both person and
> subject topics, in associations with books) and make it into=20
> a topic type,
> then make an association between this topic type and person.=20
> But again, how
> would I ensure the correct transfer between person and=20
> subject role via the
> person's books?
> 3) it also occurred to me that if the subject terms were=20
> conceived of as
> properties of the book, the transfer would be safe, and I=20
> could simply have
> these subject-properties turn out as book names in a
> "person-subject-profile" scope. However, I need the subjects=20
> to be topics in
> order to say other things about them (such as what their=20
> positions are in
> the subject term hierarchy).
>=20
> Any good ideas (and patience), anyone?
>=20
> Knut
>=20
>   =20
>=20
> _______________________________________________
> topicmapmail mailing list
> topicmapmail@infoloom.com
> http://www.infoloom.com/mailman/listinfo/topicmapmail
>=20