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

[Alexander Sigel]

Dear topic mappers,

thanks for the answers by Lars Marius Garshol, Dmitry Bogachev and
Patrick Durusau concerning my question on
how to best model associations with different modes.

This modelling should be elegant, but also display the names of the
association types as seen from the role player scopes correctly for all
modes.
I still did not figure out how to do this (See also my second posting of
2006-10-17).

I think every of my association types has two role player types (e.g.
cause and effect).
In addition, mode is some simplified logical operator which seems to be
a mix between:
* how many people claim a statement (majority or minority view)
* if the statement holds FOR ALL or only FOR SOME instances.

#1: IS: it is generally held that true FOR ALL instances
#2: MAY: it is generally held that true FOR SOME instances
#3: CLAIMED: only some claim that true (FOR SOME or FOR ALL instances??)
#4: NOT: is is generally held that NOT true (FOR ALL instances??) 

In the following, I'll go through the answers one by one:

Lars Marius Garshol 2006-10-21
------------------------------
#1
(my example)
>> (*1*) assertive (is [default])
>> fish staircases IS-BENEFICIAL-FOR (*1*) - protection of living 
>> marine resources

(lmg)
>  (1) This is just a plain statement of the fact that fish staircases
>      are beneficial for the protection of living marine resources.

Yes. A statement on a fact most would agree on.
Not disputed, and generally true (it is generally held that every
instance of fish stair cases is beneficial for the protection of living
marine resources).

#2
(my example)
>> (*2*) modal (may be [possible world])
>> raising the vat IS-INSTRUMENTAL-FOR (*2*) - cut in state deficit

(lmg)
>  (2) This statement is qualified, in the sense that the represented
>      fact is not 100% certain.
No. It is not about certainty, but rather about quantifiers: FOR SOME
instead of FOR ALL.
I'd rather put it like this:
Typically, there is not a dispute on this fact.
But the fact is not true for all cases, only for some.

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.

#3
(my example)
>> (*3*) intention (is intended or claimed to be)
>> killing of cats for experiments IS (*3*) - murder

(lmg)
>  (3) Here the statement is an claim by one group of people that many
>      other groups disagree with.
Yes, the statement is only intended (claimed) to be by some. It is a
particular viewpoint, and could be quite controversial.

#4
(my example)
>> (*4*) negation (is not)).
>> (4) Frankfurt Stock Exchange IS-UNITING-WITH (*4*) - London Stock
Exchange

(lmg)
>  (4) And this is just plain not true.
> Did I get this right? If I did, why do you want to assert (4) at all,
if it's known not to be true?

Yes, I want to represent with a statement a fact which does not hold.
It is generally believed that it does not hold.

Maybe the example #4 was not very helpful here.

Let's assume the application scenario is knowledge engineers
representing useful statements from texts
in order to create an environment in which users can explore such
statements in a knowledge network, for serendipitous learning.
One example might be a cook learning about special dietary requirements
for diabetes patients.

In this context, it seems quite useful to explicitly state e.g. common
misconceptions, or not well-known negative assertions.

In contrast to the closed world assumption, it seems we are operating
here under an open world assumption,
(stating that lack of knowledge does not imply falsity), so do well in
representing facts known to be false.

Dmitry Bogachev 2006-10-21
--------------------------
Truth values refer to the certainty [0..1] (as e.g. in fuzzy logic) we
choose to model the strength of a belief, i.e. how probable it is that
this is true.
Although modelling belief networks with numerical strength values can be
useful in some appliations, it was not what I had in mind here.
I just wanted to model the differentiation between: is regarded as true
in call cases without much dispute, is regarded as true for some cases,
is only claimed to be (and there is likely dispute), and is regarded as
not true at all.

> assertion truth with possible values: false, default false, unknown,
default true, true.
Seems to refer to a non binary logic here and could be linked to an open
world assumption.

However, my interest for the moment is limited to knowing how to
represent the modes with keeping the role type naming trick.

Patrick Durusau 2006-10-22
--------------------------
> I think the question Alexander is asking is how to handle an encounter
with a topic map that contains (4) and he wishes to say that he thinks
it is false.
In the first place I wanted to know how to best model that I know that
something is not true,
independently of someone else having claimed it to be true, or of
merging. I assume having already encountered a text stating some
negating fact, and agree on it.

Concerning merging:
Until copying into your view, every topic and assertion is scoped by the
issuer.

> "a "truth-of" assertion be defined where assertion
> (4) is a role player where an author (also a role player), in their
respective roles, asserts the assertion is false. If the truthfulness of
an assertion cannot be
> treated as a subject, it would be a serious deficiency of the
paradigm."
I would use scope (is true in someone elses view, but not in mine) or
reification (I state something about a different statement) for this
purpose.

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