& B WO 0 B ARNXMASARCOXMSÁAA WUXZ OV AB NA NAVARRE NA
v4 A74 €
we. Se T0 6 50 Bl Bl ul Su. NB DNL
LINGUISTIC CONFUSION IN SEMANTIC MODELLING
Hans-Peter Bahr
Professor Dr.-Ing., Institute for Photogrammetry and Remote Sensing
Karlsruhe University, Germany
Anita Schwender
Professor, Universidad Tecnológica Nacional - Regional Santa Fe
Santa Fe, Argentina
Commission VI
KEY WORDS: Linguistics, Concepts, Knowledge Space, System of Concepts, Semantic Net, Iconic/Symbolic Level, Paradigm
Shift
"Es genügt noch nicht, um sich einander zu verstehen, daB
man dieselben Worte gebraucht; man muB dieselben Worte
auch für dieselbe Gattung innerer Erlebnisse gebrauchen, man
muB zuletzt seine Erfahrung miteinander gemein haben"
Friedrich Nietzsche
(" Jenseits von Gut und Bóse")
ABSTRACT
The consistent development of digital photogrammetry necessarily leads to wholly new works ("paradigm shift”). Computer
assisted capture and processing of attribute data requires semantic models which demand methods different from those applied
by geometrical models. A completely new range of problem are linguistic aspects of semantic modelling. The present paper
seeks to point out common features of semantinc modelling in linguistics and image processing. This is shown by means of
a transformation concept (" real world - iconic level - symbolic level") and systematically ordered examples ("semantic nets",
"system of concepts”, "unprecise concepts”, "inexact distribution”, "inadequate methods”)
KURZFASSUNG
Die konsequente Entwicklung digitaler Photogrammetrie führt notwendigerweise zu Aufgaben völlig neuer Art ("Paradig-
masprung" ). Rechnergestützte Extraktion und Verarbeitung von Attributdaten benötigt semantische Modelle, welche andere
Methoden erfordern als geometrische Modelle. Ein ganz neuer Problemkreis in diesem Zusammenhang sind linguistische As-
pekte bei semantischer Modellierung. Die vorliegende Publikation versucht, Gemeinsamkeiten semantischer Modellierung in
Linguistik und Bildverarbeitung herauszustellen. Dies geschieht anhand eines Transformationskonzepts (" reale Welt - ikonische
Ebene - symbolische Ebene") und systematisch geordneter Beispiele (" semantisches Netz", "unscharfe Begriffe" , " unscharfe
Verteilungen" , " nicht angepaBte Methodik" ).
1 A PARADIGM SHIFT, ITS ORIGIN AND ITS by the new approach.
CONSEQUENCES Analytical photogrammetry limits modelling to just the ge-
ometry of the image. This naturally makes available only a
very narrow range of the great amount of information con-
tained in the image. Carriers of geometric information are
image coordinates. Photogrammetric models for image ge-
ometry based on the rules of analytical geometry have been
Photogrammetric image processing has undergone various very exhaustively designed and practically applied during the
transformations on its development from analogue over ana- last decades.
lytic towards digital methods.
A paradigm shift in any discipline is to be understood as the
overall, radical change of a closed set of concepts and rules
into another closed set of concepts and rules of both technical
and/or non-technical nature.
Digital photogrammetry, which stands here for a part of im-
The novel element in the step from analogue to analytical age processing by a digital computer made available beside
processing was the introduction and consistent use of coordi- geometrical image information an entirely new field of inter-
nates. This opened the possibility of modelling the geomet- est: the semantic information (sema: greek for "sign" ). "Se-
rical aspects of the real world in numerical form. mantics of the image" simply signifies " meaningful content"
in the sense of attribute data as opposed to geometrical data.
The treatment of this type of information, however, requires
not only a dramatic change of tools but also a new way of
"seeing" the image content since well established mathemat-
ical models like those applied in analytical geometry are no
longer satisfactory.
The step from analytical to digital processing is represented
by the fact that by this approach the entire analogue im-
age becomes available to the computer in numerical form.
Geometrical data are now supplemented by a scale of gray
values. The paradigm shift proper (Ackermann, 1995), how-
ever, did not occur when analogue methods were substituted
by analytical ones but when these gave way to purely digi- On the other hand, it is evident that geometry and semantics
tal processing by taking advantage of the entirely new set of cannot be separated: each object in an image has its very
possibilities with the consistent use of the possibilities offered ^ specific coordinates, but coordinates which do not designate
7
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B6. Vienna 1996