[Sigia-l] Distributed thesaurus?

[Lars Marius Garshol]

* Lars Marius Garshol
|
| Of course, the problem here is that unless the parties merging topic
| maps have had prior communication so that they use the same subject
| identifiers their topic maps will not merge. [...]

* Eric Scheid
| 
| At a pinch, one could simply use http://dictionary.com/search?q=word, 
| right? 

That's correct. It's not perfect, but in many cases it will be a lot
better than nothing.

| At least, for those words without multiple meanings ;-)

Exactly! :)

Actually, if the dictionary definitions has anchors in it for each
meaning (as it certainly ought to!) you can use it even in such cases.

| I'd like to go even further away from full scale bulk merging ...
| spidering through connections, like as envisaged with the FOAF
| network.  

You can do that with topic maps as well. In fact, this is the vision
of the person who started the whole thing (he calls it "global
knowledge interchange"). So far people have not used them this way,
they've mostly been used by corporations who are very unwilling to
share their data.

| Although, what you've described above with published subjects means
| this won't really be possible, unless those canonical sources
| reflect back locations of the various topic maps that are subscribed
| to that definition ... 

Either I misunderstand you, or it's the other way around. Typically,
as you spider around you'd find statements such as 

  "Lars Marius lives in Oslo" (from source A)
  "Oslo is in Norway" (from source B)

and if sources A and B have referred to the same subject indicator (as
they are likely to, once the OASIS GeoLang TC gets round to publishing
the indicator set for UN LOCODE) your spider will know that the
implicit statement

  "Lars Marius lives in Norway"

is also true.

There will be no canonical definitions of any of this, however. Any
definition, published by anyone, will do. So it may well be that you
can't relate the statements from A and B together before you discover
from C that the indicator A used for Oslo and the one that B used for
Oslo actually represent the same thing.

This is also how FOAF works, except that they don't have a notion of
published subject indicators. The RDF people are looking at using
those in RDF as well, but they need to agree on how first, and they
haven't yet.

In the meantime you can and do identify people in FOAF by their email
adresses, but that works exactly the same way in topic maps (I've done
a conversion FOAF->TM that does so), and for other subjects FOAF fares
rather poorly with respect to merging. (Actually, the schema is
designed so that it contains very little other than people, which is a
bit strange to me, but there may be a reason for it.)

| Perhaps if a topic X in my TM included a reference to not only the
| canonical X, but also to other X's defined in other TMs.

That is possible, actually. Topics can refer to other topics as their
subject indicators, and they can have as many subject indicators as
they want.
 
| If I had a topic map on the subject of (say) IA, in which I connect
| lots of IA subjects, some|many with the thesaurus terminology of
| BT/NT/RT ...  does the syntax of connecting A->B support B existing
| elsewhere (ie. not in my topic map).

Absolutely. That's how it works. You say (using the textual syntax
known as LTM):

  [ontopia : company = "Ontopia A/S"
   @"http://psi.ontopia.net/ontopia/#1"]

The URI given above points to the subject indicator. (Sorry about that
example, it's a bit out of date, and so should be taken with a pinch
of salt.)

If in a different topic map you then say the same thing a topic map
processor will know that this is a single subject.

| This would be useful for both outwards and inwards scope management:
| I could connect my top level concepts to a broader web-design topic
| map elsewhere, and also connect bottom level concepts to elsewhere.

Yep. This is the idea.

| Thus, if the really hard-headed nitty-gritty bamboozle-me maths
| involved in co-citation analysis is out of scope for my topic map,
| but I nonetheless refer to co-citation analysis as one way to
| improve search functionality, can I include a link to some other
| site that specialises in that subject?

Exactly. That's what you are meant to be able to do.

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
ISO SC34/WG3, OASIS GeoLang TC        <URL: http://www.garshol.priv.no >