[topicmapmail] Contextualized Topic Maps.

[Murray Altheim]

Thomas Schwotzer wrote:
> Murray Altheim wrote:
[...]
>>These are the only two places where the word "interpret" is included
>>in your analysis. I would suggest that any discussion of knowledge
>>include the context of interpretation. I.e., a common flow definition
>>of information is "that which causes a change in its receiver". For
>>information to be received by a person is one thing, but for it to
>>become knowledge, it must be interpreted. Context is certainly
>>important, but the most important context is that of the interpretation
>>itself.
> 
> I don't think so. In knowledge management there is the
> concept of tacit knowledge,. E.g. skill are tacit knowledge.
 >
> I wouldn't call it interpretation when I ride my bicycle.

You are performing an action when you ride your bicycle. I'm
not sure how that applies to this discussion. If one were to
be discussing how one learns to ride a bicycle, how one instructs
another in doing so, or representing "riding a bicycle" then we'd
be discussing the same thing. The performing of an action might
not even involve conscious thought, but (I thought) we were
discussing knowledge representation in Topic Maps.

> You are right for explicit knowledge. Consuming an external
> source is an act of interpretation.
>
>>>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.
>>
>>I think the idea of scoping topics is a bit of a non-sequitor, and
>>is perhaps a mistaken understanding of the Topic Maps paradigm.
>>In the real world as well as in Topic Maps, topics simply *exist*
> 
> 	> (in TM, Topics are proxies for their subjects). I believe what
> 
>>you really want is the ability to provide a context in which those
>>Topics play a role in an Association. And actually, Associations in
>>Topic Maps can be scoped. 
> 
> Really? As far as I know, I can scope roles in an assocation
> but not the association itself. Has it changed recently?

No change. It's always been this way. From XTM 1.0:

   "An association is a relationship between one or more topics, each
    of which plays a role as a member of that association. The roles
    a topic plays in associations are among the characteristics that
    can be assigned to it and are therefore governed by scope."
    http://www.topicmaps.org/xtm/1.0/#desc-association

>  > As to the rest of the analysis in the
> 
>>above paragraph, these are not deficiencies in the Topic Map
>>paradigm, and are all the kinds of things that can be done (and are
>>done) in Topic Map applications. 
> 
> 
> Obviously, I did not make myself clear.
> I was focussing more on the TM syntax than on the overall
> paradigm when I complained about scopes.

They are not different in this regard.

> Of course, I know that I can do whatever I want with a topic map
> application. Actually, I already do. My only point was that
> scopes are not sufficient to define any imaginable context types.
> I guess, you'd agree.

I'm not sure why either you believe that or think I would. You
can create a Topic as a proxy for any subject, so anything you
can imagine can become a Topic and therefore act as a context
or scoping agent within an ontology. And given that such a Topic
can have all sorts of other relationships within the ontology,
such scoping/contextualization can be quite rich.

>  > Other statements are again mistaken
> 
>>on the basis of Topic scoping, such as the idea that
>>
>> > 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.
>>
>>To "contextualize" a Topic in relation (association) to a given
>>subject, you'd scope those associations. You can then perform
>>any operations, such as removal, that you like. 
> 
> 
> Of course, you can add a topic for each context (type) to
> a topic map and and associate any topic that fits to the context.
> Is that what you mean? How would you handle overlapping context types?

Via multiple associations, or by adding multiple <topicRef> elements
to a Topic's <instanceOf> (same thing, really, depending on how one
models it).

> Comming back to the pizza. Medical indications and mental states
> might influence each other (if one feels sick and is also diabetic
> he/she should avoid some food). Both are context types which
> do influence each other. Such problems are usually handled with
> logical terms. How would you solve it with associations?
> Creating intersection of associated topics? Works well in this example.
> But the relation between both context types must be explicitly known.

Yes, certainly. If you want formal FOL-style inferencing, you'd create
Topics for FOL relations (like AND, OR, NOT) and use them to associate
the "pre-mixed" Topics in some fashion. This could be as complex as
the relations you're modeling. A long time ago now I did a conversion
of part of Cyc's upper ontology as a first pass at its FOL entities. A
set of those PSIs could be the basis of an inferencing engine (I've got
a portion of that hardwired into my own application).

> Of course, I can solve all these things with an application e.g.
> by defining some TOLOG statements.
> 
> But what is required to make a Topic Map Engine a bit more context
> aware? I don't have a solution. Are there classes of context types
> which deserve additional concepts? Conceptually, I would never
> propose to enhance the TM standard. It is fine as it is.
> I think of defining some kinds of context types which can be
> formalized by topics and associations. Topic Map Engines which
> are aware of these context types can perform some additional
> functions.

Yes, I think we are in agreement. TMs provide the basic structural
foundation upon which you can build ontologies, in a similar way to
how RDF provides the basis of OWL.

> An example: One of my students (Holger Hammel) combined a
> a Topic Map Engine with a geografic information system (GIS) in
> it diploma thesis. In this system location is a context type.
> The locations, coordinates, areas etc. are mostly defined by topics
> and associations. Thus, for the Topic Map Engine it was an ordinary
> Topic Map. But the LBS-TM Engine knows better and can interpret
> some special types of topic and associations as location context.
> This system can be asked for information of a given subject
> (semantical context) at a location (spatial context). The result
> is a topic map fragment which only contains information which
> are related to the subject and relevant at a the location.
> 
> Actually, the fragment does not contain any location information
> anymore. It was decontexualized.

Sounds like a very cool application. I'm certain people on this list
would be happy to have a link to information about the project.

> Sorry, it was a lengthy reply, again.

You almost tie the record!  :-)

Thanks,

Murray

......................................................................
Murray Altheim                    http://kmi.open.ac.uk/people/murray/
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK               .

     Coca-Cola's new brand of "pure" bottled water, Dasani,
     was revealed earlier this month to be tap water taken
     from the mains. Then it emerged that what the firm
     described as its "highly sophisticated purification
     process", based on Nasa spacecraft technology, was in
     fact reverse osmosis used in many modest domestic water
     purification units. Yesterday, the entire UK supply of
     Dasani was pulled off the shelves because it has been
     contaminated with bromate, a cancer-causing chemical.
     "Things get worse with Coke" -- The Guardian
     http://www.guardian.co.uk/business/story/0,3604,1174127,00.html

     Prof Steve Squyres, principal scientific investigator
     for the rover said evidence from bromide salts in the
     rock pointed to evaporation. This, he said, suggested
     a past environment where water came and went.
     "Mars rover sits on ancient beach" -- BBC News
     http://news.bbc.co.uk/1/hi/sci/tech/3560867.stm

     Martians drank Dasani!