Information Technology Reference
In-Depth Information
Chapter 3
Cognitive Information Systems
Cognitive information systems were developed on the foundation of intelligent
information systems whose purpose was not just the simple analysis of data con-
sisting in recording, processing and interpreting it, but primarily an analysis by
understanding and reasoning about the semantic contents of the processed data.
Every information system which analyses a selected type of data and information
using certain features characteristic for them keeps, in its database, some knowledge
- indispensable for executing the correct analysis - which forms the basis for gener-
ating the system's expectations as to all stages of the analysis it conducts. As a result
of combining certain features found in the analysed type of data with the expecta-
tions - generated using this knowledge - about the existing semantic data/
information contained, cognitive resonance occurs as described previously.
Cognitive information systems utilise methods which define structural reason-
ing techniques serving to match patterns [92], [112]. A system which executes
cognitive data analyses very often analyses not just information or numerical data,
but also image data. In this last case, during the analysis process, the structure of
the image being analysed is compared to the structure of the image serving as such
a pattern. This comparison is conducted using strings of derivation rules which
enable the pattern to be generated unambiguously. These rules, sometimes referred
to as productions, are established as part of a specially derived grammar, which in
turn defines a certain formal language, called an image language. An image (in-
formation) thus recognised is assigned to the class to which the pattern that repre-
sents it belongs.
Cognitive analysis utilised in cognitive information systems extremely fre-
quently follows a syntactic approach which employs functional blocks for the se-
mantic analysis and interpretation of the image [26], [30]. Pre-processing stages
completed by coding the image, approximating the shapes of analysed objects, fil-
tering and processing the image fed to the input of the system make it possible to
obtain a new representation of the image presented as hierarchical structures of the
semantic tree and subsequent steps of deriving this representation from the initial
symbol of the grammar [30], [73].
During image data pre-processing, an intelligent cognitive recognition system
must (in most cases) segment the image, identify picture primitives and also de-
termine relations between them.
The classification proper (as well as the machine perception) consists in recog-
nising whether the specific representation of the input image belongs to the class
of images generated by the formal language defined by one of the grammars that
can be introduced - a sequential, a tree or a graph grammar - which are used for
recognition processes executed during the syntactic analysis performed by the
system [30], [73], [92].
