[topicmapmail] Re: How to best model relations with mode (assertive, modal, intention, negation) with topic maps?

Wed Oct 25 12:57:42 EDT 2006

Dear Jack (and dear topic mappers):

you had kindly replied with some remarks sent 2006-10-23 to my question on how to best model some assertions with topic maps.

Here is my reply [AS]:

[AS]
>> E.g. "cancer cells IS_IN_PLACE_OF [2] lungs"
>> means that lungs MAY INCLUDE cancer cells, whereas "cancer cells 
>> IS_IN_PLACE_OF [1] lungs"
>> would mean that all lungs include cancer cells.

1. PURPOSE
-----------
[JP]
What is the purpose or intention or goal of a topic map containing assertions of the type "cancer cells IS_IN_PLACE_OF[2]lungs"?
Would this be a "tutorial" topic map, a "diagnostic" topic map, an index of information resources about cancer?

[AS]
The right person to ask about the rationale or usage scenario of such an ICLO-based system is Prof. Dr. Winfried Schmitz-Esser [SE],
the originator of this idea.
I am "only" counseling here from the side of topic maps and knowledge organization.
So I will restrict myself to the question if the requirements are as set in [1], how can topic maps aid in realizing such a system.

Below you find his comments to your question, together with my comments.

[SE]
ICLO is designed as a general, open model.
As you say, I could be most useful for tutorials (see the example in fig. 17, p.96 [1]
[Argument "cell motility" as modelled from knowledge contained in note on insulin receptor substrate 2 published in Cancer Gene Card]).

But it may be used for diagnostic, research, and even knowledge generation purposes as well.
The aim is that any piece of knowledge can be represented in this ontology, irrespective of domain or language,
provided the knowledge comes as natural language text in one of the languages implemented on the tm-based KOS [Knowledge Organization System]. (p.74). 

[AS]
We have reason to believe that there is room for a system for human exploration of knowledge networks and are looking for interesting application scenarios.
I do not see it for diagnostic purposes, but as a more comprehensible representation of statements about the domain
than represented as text in scattered documents, inspiring insight.

2. APPROPRIATE MODELLING
------------------------
[JP]
Is location or containment the rigt way to model populations of cancer cells existing in particular organs or tissues?
What about mereonomy here?

[AS]
A mereonymy expresses part-whole relations (partonymy). You could not recognize this from my email or the particular relation type name,
But this particular relation type is a subtype of a part-whole relation, so this is perfect mereonymy relationship type.
I think, this is an appropriate modelling, but maybe the relationship type name could be improved.

[SE]
ICLO RelClass2-Type 3.1 Location (part/whole) is conceived as a mereonomy.
The definition (p.100) says: "A is the part as by location"/B is the whole as by location".
The relation (association type) is one of four subtypes of RelClass2-Type 3 Partitive.
The graph on p.74 is a model of the knowledge expressed in an English language text encountered on the Cancer Gene Card knowledge base.
In a way, the statements underlying the graph form a formalized model from the respective universe of discourse. 

[JP]
stresses the importance of modelling assertions properly,
and comments on the need of more specification, illustrating this on the example of leukemia.

[SE]
ICLO sees no limits as to specificity or generality of the arguments (represented by base names/alternative names).
It all depends on the relations. Whether a practical, useful level of expressiveness can be achieved on the basis of the proposed "Arsenal" of 23 defined Class-2 relations (plus some optional) remains an open question. 

[AS]
We found the 23 relationship types quite useful for some purposes, but would not claim to have found a complete set for all applications.
In addition, there would be application-specific relations.
However, for the moment being, we are quite content with organizing knowledge (statements found in texts) based on those basic relationship types.
In my TMRA06 tutorial on knowledge organization with topic maps I had stressed how important it is to control not only the topic types,
but also the relationship types against the arbitrary proliferation of semantic relations. Semantic relation types are defined by their role types,
which are bound to categories, which can be controlled.

[JP]
> I guess that for purposes of making an example of a 
> particular assertion, "cancer cells is_in_place_of lungs" is 
> a reasonable statement

[AS]
Yes, it exists just as one example statement, nothing more.
A better interpretation of the statement is: 
"cancer cells IS_IN_PLACE_OF [2] lungs"
Lungs MAY BE the mereonymic whole in the sense of a location for cancer cells, which MAY BE the meronymic part.

[JP]
for me, it raised more issues than it satisfied.

[AS]
Sure. Maybe we have to look for a better example.
My interest in posting to this list for the moment was only how to best model such a kind of modal modifier with topic maps,
irrespective of the nature of a particular example.
IS vs. NOT would be sufficient to know.

[JP]
If reading such a map, I'd want to know more, and that curiosity plays to a particular passion/concern,
that being the use of topic maps for modeling important universes of discourse. 

[AS]
In an explorative environment your curiousity to interact with semantic knowledge networks is most welcome.
The ICLO modelling restricts itself to some forms of crude but useful modelling which are better than typical environments of today.
Correct and full comprehensive knowledge representation is not the primary aim.

[JP]
I don't see modeling as trivial. Far too many choices to make if one is to craft a topic map that goes beyond ontologies to federate human discourse.
modeling assertions properly (whatever that means) is an important role for topic mappers

[AS]
True. Neither have we found the holy grail ...
All we try is to find a reasonable simple system to cope with current deficiencies.
Suggestions for improvements are welcome.

[JP]
does this topic map specify which kind of "cancer cells"? (...)

[AS]
It is not a diagnostic map for leukemia, but records only the fact that some text
implied as commensense knowledge a location part/whole relationship type between two phrases standing for such concepts.
You may add more facts in a very narrow domain, which we did not do in this.

[JP]
I don't mean to drive the example to death, but it did catch my eye
since I saw it as an unusual (whatever that means) way 
to model modal assertions in a particular domain.

[AS]
Abstract from the domain for a moment.
We explicitly do not want to model modal logics (as I wrote before), because this opens a complexity we are not after.
We just want to discern between very simple, different interpretations of associations.

[JP]
I suspect that modeling modal logic is way larger a subject than can be 
handled in a few emails. Peirce, I am to understand, never 
did finish his gamma existential graphs, which are said to 
model modal logic.

[AS]
Right you are ...

-------------
[1]
Schmitz-Esser, W. & Sigel, A. (2006): Introducing Terminology-based Ontologies. Papers and Materials presented by the authors at the workshop "Introducing Terminology-based Ontologies" (Poli/Schmitz-Esser/Sigel) at the 9th International Conference of the International Society for Knowledge Organization (ISKO), Vienna, Austria, July 6th, 2006. 130 pages. Published electronically on E-LIS (E-prints in Library and Information Science, http://eprints.rclis.org/archive/00006612/), 2006-07-14

Best regards
alex
---
Alexander Sigel, M.A., Researcher in Semantic Knowledge Networking
Univ. of Cologne, Dept. of Information Systems & Information Management
http://www.wim.uni-koeln.de/19.0.html
sigel at wim.uni-koeln.de, +49 221 470-5322

> -----Ursprüngliche Nachricht-----
> Von: topicmapmail-bounces at infoloom.com 
> [mailto:topicmapmail-bounces at infoloom.com] Im Auftrag von Jack Park
> Gesendet: Montag, 23. Oktober 2006 15:23
> An: topicmapmail at infoloom.com
> Betreff: Re: [topicmapmail] Re: How to best model relations 
> with mode(assertive, modal, intention, negation) with topic maps?