Dear Maximilian,
Thank you very much for your detailed answer! Makes things much clearer.
Let me comment again:
On 9/1/2016 12:45 πμ, Maximilian Schich wrote:
Dear Wolfgang, Martin, and all,
My Edge contribution does not put into question the CIDOC-CRM, or the
hermeneutic circle in general. The argument is that quantitative
measurement accelerates the process and extends beyond the reach of
qualitative inquiry.
I am glad you don't put into question the CRM :-) .
The hermeneutic circle is often confused with the argument structure
itself. I do not believe we learn very much from the hermeneutic circle
(see also: Doerr, M., Kritsotaki, A., & Boutsika, A. (2011). Factual
argumentation - a core model for assertions making
<http://dl.acm.org/citation.cfm?id=1921615>. /Journal on Computing and
Cultural Heritage (JOCCH) /, /3/(3), 34, New York, NY, USA : ACM) . I
think it distracts the focus from
the actual logic of inference making in sciences and scholarship. It
describes the accidental, the surface, rather than the substantial. I'd
see it as an idealization of the actual complex interaction patterns you
describe, as if a small group of researchers would spiral in their own
soup, without giving insight in the scholarly logic. So, in both ways,
there is not much more to learn from it.
The CRM has nothing to do with the hermeneutic circle. It describes a
part of the factual world cultural-historical and scientific
investigations are interested in and can easily decide on shared notions
of identity, which enables information integration. In a hermenautic
circle, researchers may like to modify one or more CRM instances graphs.
The CRM does not prescribe research in any way.
If you replace "measurement" with "observation", I am much more on your
line. Sciences such as geology and
life sciences are full of observations, which hardly can be called
measurements, such as the observation of species
occurrences. I was more objecting against the often silent assumption
that "measurement" provides per se objectivity
and is superior to observation.
I would further distinguish between quantitative measurement and systematic
observation that achieves some sense of coverage of phenomena. When it
comes to scientific break-through, the most rare thought may win, as
Paul Feyerabend describes in "Against Methods". Quantitative analysis
may project the ineffective or irrelevant.
Systematic observation indeed accelerates the process. What is effective
systematics, is a highly complex in its own.
Indeed, in CRM-SIG we propagate systematic observation. If we succeed,
is another question ;-) .
For a more extended account with figures, please take a look at
Maximilian Schich: Figuring out Art History. DAH-Journal 2 (2016) [to
appear]
Preprint: http://arxiv.org/abs/1512.03301 (22 Oct 2015)
I have read your article with interest. I completely follow you, that
research in collective behavior has to be treated as you describe, and
that new insight will be gained from that, and that disciplinary
narrow-mindedness is an obstacle.
I note however, that the data collection you silently assume, is based
on individual facts, for which you need other processes to validate
them. We must not confuse the parameter selection for statistical
studies of any kind with ontologies in the proper sense, which provide
the overall causal model of the world from which we select parameters.
These parameter sets may appear as "data models" as you describe. In the
"ACGT" research project, the Consortium could show that all datamodels
of clinical studies in cancer research on chemotherapy could be
described as views of one global ontology, a "superdatamodel", which is
more and more stabilized around scientific insight on the human body.
From the examples you gave in your paper for instance, I do not see
anything that would not be based on a CRM compatible data collection.
Therefore, I do not follow your figure 5, and your claims about
datamodels. I do not see how you could feedback the working hypothesis
expressed in the datamodel statistically without a ground theory. For
the latter, a "speed up" would be just the opposite of an advancement of
science. From your paper, I do not see any argument how this should work
out. Please give more details on that! Note, that physics over centuries
stabilizes more and more around fundamental laws, which more and more
increase precision of predictions.
In general, I'd be more cautious with global claims such as "the actual
history of all made things" or "a perspective
for a systematic science of art and culture" ;-) .
Please let me also answer your questions briefly...
Martin's questions:
* Like glaciers, frozen models still move and are often beautiful
and vital as a "reference". ;)
Well, I hope you don't see the CRM like that. For me the CRM is a domain
theory as for instance Newton's mechanics. If you have a clearly defined
scope and precision requirement, it gives the correct answers. It is not
a question of beauty or arbitrary agreement. If
it does not give the correct answers, you change it. It is veryfied
empirically. If you go beyond its initial scope, you may need to modify
it it, such as theory of Relativity. If you seek different kinds of
answers, you find another model, such as Thermodynamics.
If there are competing models for the same questions, either one wins by
better answers, or you simply merge them.
We have merged the CRM with the ABC Harmony Model and with FRBR. That is
not a moving glacier. We will continue to merge the CRM with any model
that will provide better answers in the same scope. See above about
datamodels.
* The initial hypotheses may be established in a traditional way:
Facebook for example has updated their traditional gender-model
from male/female to a list of 250 genders, emerging from user
input, characterized by a frequency distribution and temporal
dynamics that can only be measured in a quantitative way. The
emerging ENRON email structure versus ENRON's defined corporate
hierarchy is another good example for the need of quantification.
We should not confuse data models with terminology. Terminology is
justified by intensional definitions, which are based on relationships,
which are more fundamental and much less. Terminology has a different
epistemological role than datamodels.You cannot create "data" with
terms, only with relationsships. Terms appear in data as values.+ Have
we understood what 250 genders should be? When you measure the
distribution and dynamics, what sort of things do you learn from that?
Terminology is typically fluent.
*
* Yes, after measurement, disproving hypotheses, and finding a way
out of Uri Alon's "cloud of uncertainty" (aka science), the loop
needs to be closed (by engineering). Therefore, the CRM-SIG will
be even more important than before (doing both science and
engineering).
* I use the word "measurement" in the sense of Max Planck, who
claimed that any observation of the "real world" is subject to
measurement bias, either due to imperfections in our tools or our
own sensory organs.
Sure.
*
* Quantity is not indicating quality per se. But quantification can
reveal hidden quality as "more is different".
sure
* There is no confusion: Cultural research is part of the cultural
process itself.
I think there is still confusion. Of course is cultural research part of
the cultural process, but not because it researches culture, but because
all human activities are part of culture. Studying the research process
reveals the research culture, but not the culture researched. It may
help improve the research process, and by that indirectly improve the
subject matter. Scientific insight comes from discrimination, not from
declaring all to be the same.
*
* Yes, physicists are aiming to improve physics by studying
interaction patterns between physicists => Sinatra et al. Nature
Physics 11, 791-796 (2015) doi:10.1038/nphys3494 (cf. final paragraph)
Sorry for my floppy expression! I meant you will not find a new law of
particle physics by studying the interaction patterns between physicists.
Wolfgang's questions:
* Of course, data comes from databases old and new. In fact,
analyzing old datasets most interesting, as their data models were
usually formulated decades ago, without knowing the emerging
complex patterns that result from "local activity" by curators and
the heterogeneity of granular data collected over time.
Still I'd need concrete examples how activity patterns of curators would
lead to question datamodels. As a global statement, it makes no sense to
me, I will not exclude that there may be a particular effect some times.
* The data and method of our Science paper is published in the
Supporting Online Material (free access to the Science website via
www.cultsci.net). This allows for reuse and feedback of conceptual
ideas by others. I assume all steps of the hypercycle will have
their own publication stream, feeding into following steps.
* Our conceptual reference models are out of sync with (a) a large
number of databases with tens of thousands of entity and property
types, and (b) massive amounts of data where the entire
ontological structure is hidden (for example in plain
tagging/category systems). In both cases, quantification is
essential to model the emerging structure and dynamics, and
eventually update our conceptual models.
Well, our research in mapping thousands of database fields to the CRM
neither shows an "out of sync", nor that "emergent semantics", which may
play well for taxonomic systems, can apply to database design, and I
cannot understand how the "dynamics" help eventually update conceptual
models. The problem being, that the database surface structure is quite
different from the mental models behind (see Fauconnier, "The Way We
Think"). But if you can show us a nice application that works, I'll be
more than happy to use your methods!
In sum, I see a world with much more analytical structure of very
different, in which your research covers a certain valuable area, as
mine does at another edge, but I would be hesitant with too global
claims, such as " perspective for a systematic science of art and
culture" ;-) .
All the best,
Martin
In sum, all stages of the hermeneutic hypercycle are essential.
Quantification will play an important part. But this does not mean
traditional ontology engineering will go away.
Best regards,
Max
*Dr. Maximilian Schich*
Associate Professor, Arts & Technology
Founding member, The Edith O'Donnell Institute of Art History
*/The University of Texas at Dallas/*
800 West Campbell Road, AT10
Richardson, Texas 75080 – USA
US phone: +1-214-673-3051
EU phone: +49-179-667-8041
www.utdallas.edu/atec/schich/ <http://www.utdallas.edu/atec/schich/>
www.schich.info <http://www.schich.info>
www.cultsci.net <http://www.cultsci.net>
Current location: Dallas, Texas
On 2016-01-08 8:58 , martin wrote:
Dear All,
Just to add to Wolfgang's remark:
The CRM is in no point a product of a priori intuition, but
exclusively based on empirical study of
database use and interpretation, and a continuous feed back to
systematic updates of the CRM.
More flexible mapping mechanisms and semantic Web technologies also
enable the systematic update
of the databases to new releases. The CRM, as ISO standard, is not
"frozen", but has the regular update
cycle of 5 years, which CRM SIG extensively uses.
How ontological relations can emerge from quantitative measurements
is black magic to me:
All quantitative measurement requires an a priori hypotheses, and
competing hypotheses will reveal better
or worse agreement with reality. So far sciences appear to me to
work. So, what are the initial hypotheses
about such patterns? Or is there again an ontology engineering step
after the measurement?
I agree that the real ontological patterns are often not what expert
intuition would suggest in the first place.
This is our common experience. However once found to be operational,
they must be compatible with
scientific argumentation and expert can confirm. I agree with
Maximilian that data structures must be based on
empirical research, but "measurement"?
The "quantitative" argument is equally puzzling to me. Is quantity
now indicating quality? Aren't we
here confusing the sociology of doing cultural research and the
evolution of knowledge with nature
of the subject matter and the structure and logic of the scholarly
argument? Would anybody reasonably
try to improve the science of physics by studying interaction
patterns between physicists???
All the best,
martin
On 8/1/2016 9:12 πμ, Wolfgang Schmidle wrote:
Dear All,
Let me quote from fellow list member Maximilian Schich's critique of
database models and CRM:
"Over decades, database models, to embody the underlying worldview,
were mostly established using formal logic and a priori expert
intuition. Database curators were subsequently used to collect vast
numbers of specific observations, enabling further traditional
research, while failing to feed back systematic updates into the
underlying database models.
As a consequence, "conceptual reference models" are frozen,
sometimes as ISO standards, and out of sync with the non-intuitive
complex patterns that would emerge from large numbers of specific
observations by quantitative measurement. A systematic data science
of art and culture is now closing the loop using quantification,
computation, and visualization in addition."
http://edge.org/response-detail/26784
Max, let me start by asking where the data underlying visualisations
such as https://www.youtube.com/watch?v=4gIhRkCcD4U is supposed to
come from, if not an old-fashioned database? How did you feed the
"non-intuitive complex patterns" emerging in this visualisation back
into the database or somewhere else? And why do you think CRM is out
of snyc with this?
Thanks
Wolfgang
_______________________________________________
Crm-sig mailing list
[email protected]
http://lists.ics.forth.gr/mailman/listinfo/crm-sig
--
--------------------------------------------------------------
Dr. Martin Doerr | Vox:+30(2810)391625 |
Research Director | Fax:+30(2810)391638 |
| Email:[email protected] |
|
Center for Cultural Informatics |
Information Systems Laboratory |
Institute of Computer Science |
Foundation for Research and Technology - Hellas (FORTH) |
|
N.Plastira 100, Vassilika Vouton, |
GR70013 Heraklion,Crete,Greece |
|
Web-site:http://www.ics.forth.gr/isl |
--------------------------------------------------------------
