[topicmapmail] Agents and Topic Maps

René Schmitt edelholz@xterminator.studfb.unibw-muenchen.de
Fri, 28 Feb 2003 22:36:25 +0100


> I am currently developing a multiagent application, which uses Topic
> Maps. My agents have a predefined (base) Topic Map for a specific
> domain. Each agent tries to classify various ressources (bookmarks,
> documents etc.) to the defined topics, and adds links to the ressources
> as occurrences. This agent produces a TM, which can be useful for the
> user (local use) and tells a lot about the user and his interests in
> specific topics. An agent can go to various hosts building local TMs.
> Two (or more) agents meeting, can merge the maps to get one TM, or go to
> a central host, where a complete TM (with all occurences from all
> agents) is created (merged). This map may help people, to find others,
> which are interested in the same topics, and makes (local) ressources
> available for all. Agents are used to create different visualizations of
> the map (xhtml with xslt, java based with tmnav), too.
> I am not sure, if this approach makes sense, but (as far as I know) it
> is a new approach, so it seems worth trying.

This is exactly what I search for. Your approach seems very promising to me.
As far as I understand, you´re going to build up a multi-agent application
for LAN purpose (if it works you´ll probably want to build an internet
solution).

But few thing are not clear.
How does your agent verbalise it´s objectives?
It has to communicate with other agents, does it use KIF or other languages?

For example:
It´s starts to search for DVD´s with Tom Hanks as leading actor in a certain
environment. In this environment exists a global Topic Map with all
information
and resources provided.
Now the agent has a starting (base) TM in it´s data package.
How does it tell the server what it searchs for?
Is the base TM empty and will the agent adds topics and occurrences if some
fitting sources are found?
Or is the base TM equal to the global topic map and the agent is restricting
it
by using scopes, facets, inference or whatever to determine what data
is irrelevant and to be deleted?

Nice Weekend to you all and besides
Thanks for your help
René Schmitt