[topicmapmail] Contextualized Topic Maps.

[Thomas Schwotzer]

> My own view is that, somewhat akin to the old biology mantra: "it's all 
> in the context", this is an interesting thread.  Francis Heylighen [1] 
> (couldn't get the site to open, so couldn't select appropriate papers) 
> has written about "learning webs" where a hebbian learning algorithm 
> assigns and massages weights on links according to click counts. Lots of 
> people study the notion of "context". When you think about mapping that 
> into topic maps, the <scope> tag comes into play, and more.
> 
> Don't truncate this thread.
> Cheers,
> Jack
> 
> [1] http://pespmc1.vub.ac.be/papers/PapersFH.html

Context is definitly an issue. Whenever topic maps are
used to structure something that could (even only remotly) called
knowledge context is required.

Topic Maps — Data Model:
"Topic maps are abstract structures that can encode knowledge and 
connect this encoded knowledge to relevant information resources."

What is knowledge? I'm afraid that would be out of scope
of this mailing list. Just some short statements about it:

Knowledge is something that helps to solve a problem (not a
citation but it was something like this). Explicit knowledge
is knowledge that was written into a document (externalized)
and exists outside human brain.
(Nonaka/Takechi: "The Knowledge-creating company")

Thus, information resources of TM could be intepreted as explicit
knowledge.

Moreover, a couple of authors (beside Nonaka/Takeuchi) have stressed
that every kind of knowledge requires a context.

There are several kinds of context, e.g. semantics.
I think, topics and associations can be interpreted as
(semantical) context.

Topic map authors attach information resources (maybe explicit
knowledge) to topics. In other words, they put the knowledge into a
(semantical) context.

But this is done in a context, again. Authors do their work
in a mental state at a dedicated place. Thus, the assignment
of an information resource to a subject is done in a context.
Whereas the semantic context can be defined by topic maps, the
other kind of context isn't usually defined explicitly.
What's a pitty. Because, this context has major impacts on usability
of the topic map:

E.g. I trust Ontopias opera topic map because I'm pretty sure
that Steve Pepper is familiar with the topic opera. I wouldn't
trust a topic map dealing with motorbikes issued by Ontopia.
Sorry.

Thus, people who issued or structured knowledge are part of
the context themselves. Whenever I consult a topic map
I'm aware of the topic map authors (explicitly or implicitly)
if they are known. But even the fact that authors are
anonymous is an information which becomes part of my context.
This is sometimes called "background context".

Moreover, location is a kind of context. If I'm in the opera
I'm not very interested in news on topic maps. Thus, the
place where I am and also the mental situation influences
my view on information resources and the ability to use
information resources as knowledge.

To point it out: In my eyes depending an information resource can
be knowledge or not depending on the context. Right, it's all in
the context. And context is in a static flux.

Currently, topic maps can be used to define semantical context.
Such a semantical context is just a kind of a view, like a
database view.

There are more context type and different context types do
influence each other. E.g. the location were I am can influence
my interest on topics.

Context types I know are:
- persons (or groups of persons (knowledge nodes (Bonifacio)))
- mental state
- location (phyical or logical - beeing in a Internet chat rooms might
be perceived for somebody as beeing in a bar)
- time (daytime, working time etc.)
- language
- interest (Dewey: interest bias)

Comming back to the pizza example: A couple of context type
were involved:
Mental state: hungry
Medical indications: nothing / diabetic / ...
Food pattern: vegetarian / diet ...
Location: city / countryside / moon / ...

As far as I can see there are only a few approaches to cover
more than one or two context types and relations and interactions
between them (e.g. Manos Theodorakis: Contextualization)

Very interesting issue. But stop this thread if you'd think it's
not appropriate for this mailing list.

BTW, I'm going to write my PhD thesis about it.
(working title: "Context based knowledge flows in mobile
spontaneous networks").

I hope I can generalize existing context models and can map it
to Topic Maps and design a distributed TM-Systems which
exchanges Topic Map Fragments based on context via Bluetooth.
That's the plan. I hope I finish it this year.

I'm convinced that the current scope concept isn't sufficient
to cover all types of context. E.g. topics and associations cannot
be scoped. It would be useful, e.g. if a want to get a Topic Map
that only contains concepts regarding to dedicated place.
With scopes I cannot delete any topic. I can invalidate any
topic characteristic. But the topic itself remains in the
map. Scopes, in the current version, can be used to create
a view on a topic map but not to contextualize it which would
include removing topic map items. I think topic map fragmentation
can be technical means to contextualize topic maps.

Sorry, for this lengthy statement.

Cheers, Thomas

-- 
Thomas Schwotzer
University of Technology Berlin
Intelligent Networks and Management of Distributed Systems
http://ivs.tu-berlin.de/Schwotzer
fon: +49-30-314-79834 fax: +49-30-314-24573