Sent: Friday, January 14, 2011 5:50 AM
Subject: Re: [ontolog-forum] Ontology of Rough Sets
Tara,
I don’t recall any specific
reference on rough sets – I think I first read something in the IEEE
suite of pubs, maybe ten years ago.
Rough sets should be fairly easy to
implement in SQL, so it would be interesting to know what components you were
considering, and whether they met your needs.
If any reference info comes back to me,
I’ll post it.
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx]
On Behalf Of Tara Athan
Sent: Thursday, January 13, 2011
3:31 PM
To: Rich Cooper
Cc: '[ontolog-forum] '
Subject: Re: [ontolog-forum]
Ontology of Rough Sets
Rich Cooper wrote:
Hi
Tara,
Are you are visualizing the slice of an
orange while I’m visualizing the remainder of the orange with the slice
removed?
Must be, because I don't
catch how the application you outline below fits the mathematical theory of
rough sets that I have read about.
Can you provide a reference?
Without that visualization shift, I
can’t fit my imaging into your response “a rough set
would be a sub-lattice with top node of the upper approximation and bottom node
being the lower approximation”.
My view of a rough sets application is
that we have many overlapping ways of sensing objects with instruments (blood
tests, MRI …) that can detect and report on what was found. The
trajectory, or process, (e.g., life cycle of a disease) of the said object
shows up in different instruments at different phases of its existence.
Before computers, statisticians could only dismiss outliers, but with
computers in everything, we can look at the outliers as valid data points.
There must be subtleties yet unmined in those signatures.
Rough sets provide a way of modeling that
kind of dynamic signal and differencing. Potentially they can help
explain nuances of the data that weren’t detectable with less analysis.
Sensors are often modeled as
probabilistically detecting or not detecting objects, but remember that sensors
only detect COLUMNS, of object properties - attributes. So the
probabilistic view may not make practical analysis of data outliers.
So I would still like to see an ontology
of rough sets, if there is one available.
Thanks,
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
I'm more inclined to think of a lattice of crisp sets,
and a rough set would be a sub-lattice with top node of the upper approximation
and bottom node being the lower approximation.
One of the interesting things, to me, of this set theory is two orthogonal
partial orders:
1. the usual containment (U1 (upper approx of set 1) contained in U2, L1
contained in L2) -> RSet1 contained in RSet2
2. roughness (U1 contained in U3, L3 contained in L1) -> RSet1 is rougher
than RSet3
And actually this was why I started to look into the theory in the first place-
because I was considering a domain with more than one partial order - graphs.
Tara
Rich Cooper wrote:
Tara, John,
Also, the lower and upper bounds on a rough set parallels some of the
lattice properties John has suggested for ontologies. Would it be accurate
to characterize rough sets in that way - as a specification of the GLB and
LUB in a lattice of rough sets, based on data collected in the rough sets so
far?
Musingly,
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
-----Original Message-----
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Tara Athan
Sent: Thursday, January 13, 2011 10:53 AM
To: [ontolog-forum]
Subject: Re: [ontolog-forum] Ontology of Rough Sets
OK, let's be careful.
There is a mathematical set theory called "rough set theory", that was
introduced in the 1980's
PAWLAK, Z. 1982. ROUGH SETS. International Journal of Computer &
Information Sciences, 11(5), pp.341-356.
and has been studied quite a bit since. Here are some more recent articles:
PAWLAK, Z. and A. SKOWRON. 2007. Rudiments of rough sets. Information
Sciences, 177(1), pp.3-27.
PAWLAK, Z. and A. SKOWRON. 2007. Rough sets and Boolean reasoning.
Information Sciences, 177(1), pp.41-73.
PAWLAK, Z. and A. SKOWRON. 2007. Rough sets: Some extensions.
Information Sciences, 177(1), pp.28-40.
It might be a good idea to emphasize at this point that sets and
concepts are not the same thing. A set is determined by its membership,
a concept is determined by its definition. The extension of a concept is
a set, but the same concept can have extensions with different
membership in different contexts, such as at different times (this may
be an abuse of the term "context", I'm not sure.) However, if we,
implicitly or explicitly, specify the context so that a concept has a
unique extension, then we can work with sets through their defining
concepts.
Rough set theory uses the notion of "indiscernability", where there is
an attribute which can be measured to some precision or "granularity"
and this precision does not fully describe reality. The universe of
discourse is clustered into groups that indiscernible with respect to a
set of attributes.
"Crisp" sets consist of unions of these clusters. But it is also
possible to have sets, as determined by listing their membership, which
cannot be defined precisely by specifying regions in the quality spaces
of the available attributes because there are indiscernible clusters
that fall partly within and partly outside of the set.
Such sets are rough sets, and they have two approximations:
an lower approximation, which consists of the greatest crisp subset,
a upper approximation, which consists of the smallest crisp superset.
Given only the available attributes of an individual x, and the upper
and lower approximations of a rough set, then the question "Is x a
member of rough set B" has three possible answers, definitely yes,
definitely no, or maybe. Thus rough sets intuitively carry a 3- valued
logic, although the mathematics is more nuanced - see
POLKOWSKI, L. 2003. A note on 3-valued rough logic accepting decision
rules. Fundam. Inf., 61(1), pp.37-45.
So we can also wonder about the extension of the "rough" notion to
ontologies.
That approach has been explored:
INUKAI, Y., A. GEHRMANN, Y. NAGAI and S. ISHIZU. 2007. ROUGH SET THEORY
USING SIMILARITY OF OBJECTS DESCRIBED BY ONTOLOGY [online]. [Accessed
Jan 13, 2011]. Available from:
http://journals.isss.org/index.php/proceedings51st/article/viewFile/504/294.
Or we may ask a different but related question: Instead of an atomic
concept being a Boolean unary predicate, what if it is a 3-valued unary
predicate? I'm not sure if there are published papers addressing this
question.
If there are other perspectives on this issue, it would be interesting
to hear about it from other list members.
Tara
John F. Sowa wrote:
Folks,
We have to make a clear distinction between ontology and the tools,
languages, logics, and reasoning methods used with any ontology.
The subject line of this thread could be very misleading.
Ever since Aristotle, categories and hierarchies of categories
have been useful for ontology -- primarily because the study
of existence leads to a study of what kinds of things exist.
A's syllogisms and his method of definition by genus and
differentiae have also been useful. But many people (starting
with Aristotle himself) have noted that prototypes rather
than strict definitions are better for some applications.
In general, there is *no* specific logic or reasoning method
that is either essential or irrelevant to ontology. The
choice depends entirely on specific applications -- or even
on very narrow questions or problems about an application.
Re rough sets: This is an important topic, but I would be
very cautious about any way of thinking that combines the
word 'ontology' with any particular reasoning method.
That is in fact why I have been unhappy with the phrase
"Web Ontology Language" used as a scrambled acronym for OWL.
It suggests to many novices, that the word 'the' belongs in
front of that phrase -- but that idea is hopelessly misguided.
Ontology is *orthogonal* to any and all versions of logic,
reasoning methods, and implementation tools. Any specific
language or tool tends to channel thinking into certain paths
that may be useful for some applications. But ways of thinking
that are specialized for one kind of application can often be
inappropriate or even misguided for other applications.
John
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
--
Tara Athan
Owner, Athan Ecological Reconciliation Services
tara_athan at alt2is.com
707-272-2115 (cell, preferred)
707-485-1198 (office)
249 W. Gobbi St. #A
Ukiah, CA 95482
--
Tara Athan
Owner, Athan Ecological Reconciliation Services
tara_athan at alt2is.com
707-272-2115 (cell, preferred)
707-485-1198 (office)
249 W. Gobbi St. #A
Ukiah, CA 95482
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx