Information Technology Reference
In-Depth Information
The most recent studies of intelligent information systems indicate that only
recognising the analysed image is no longer sufficient, because researchers in-
creasingly frequently propose to employ these systems also for the automatic,
computer understanding of the image. This applies in particular to image data con-
taining layers of semantics. Such image data includes e.g. medical images. In or-
der to enable reasoning for a selected class of patterns, artificial intelligence tech-
niques are used. These techniques, apart from the simple recognition of the image
identified for analysing, can also extract significant semantic information which
facilitates its semantic interpretation, i.e. its full understanding.
This process applies only to cognitive information systems and is much more
complex than just recognition, as the information flow in this case is clearly two-
way [87], [92], [98], [105]. In this model, the stream of empirical data contained
in the subsystem whose job it is to record and analyse the image interferes with
the stream of expectations
generated. A certain type of interference must occur be-
tween the stream of expectations generated by the specified hypothetical
meaning
of the image and the stream of data obtained by analysing the image currently un-
der consideration. This interference means that some coincidences (of expecta-
tions and features found in the image) become more important, while others (both
consistent and inconsistent) lose importance. This interference, which in conse-
quence leads to cognitive resonance, confirms one of the possible hypotheses (in
the case of an image whose contents can be understood), or justifies a statement
that there is an irreparable inconsistency between the image currently perceived
and all gnostic hypotheses which have an understandable interpretation. The
second case means that the attempt to automatically understand the image has
failed [96].
Thus the operation of cognitive information systems hinges on cognitive reso-
nance which characterises only these systems and distinguishes them from other
information systems. The use of such systems may be varied, as today's science
offers wide-ranging opportunities for their application. The best applications for
cognitive information systems are currently found in medicine, as more and more
pathological entities are identified in disease processes of individual human organs
and the capability to detect and recognise these entities keeps improving. Medical
images are among the most varied forms of data and have extremely deep and sig-
nificant meaning interpretations. Cognitive information systems can certainly also
help in many other fields of science and everyday life if an attempt is made to add
the process of understanding the analysed information/data to intelligent informa-
tion systems in the fields of economics, marketing, management, logistics, defence
and transport.
3.1 General Classification of Cognitive Information Systems
Cognitive analysis based on processes of learning about and understanding the
studied phenomenon has become an opportunity for developing intelligent infor-
mation systems almost in every field in which data is currently analysed. The
general class of information systems using cognitive analysis for semantic inter-
pretation and reasoning is referred to as cognitive categorisation systems, among
