AUTOMATIC INFORMATION PROCESSING AND

UNDERSTANDING IN COGNITIVE BUSINESS SYSTEMS

Ryszard Tadeusiewicz, Marek R. Ogiela

AGH University of Science and Technology, Institute of Automatics, 30 Mickiewicza Ave. 30-059 Krakow, Poland

Lidia Ogiela

AGH University of Science and Technology, Faculty of Management, 30 Mickiewicza Ave. 30-059 Krakow, Poland

Keywords: Automatic understanding systems (AUS), Data analysis, Understanding-based information systems,

Cognitive business systems, Cognitive analysis.

Abstract: The concept of new generation in area of information systems is automatic understanding systems (AUS) to

the attention of the computer sciences community as a new possibility for the systems analysis and design.

The novelty of this new idea is in the previously used method of automatic understanding in the area of

medical image analysis, classification and interpretation, to a more general and needed area of systems

analysis. The concept of the AUS systems approach is, in essence, different from other approaches such as,

for example, those based on neural networks, pattern analysis, image interpretation or machine learning.

AUS enables the determination of the meaning of analysed data, both numeric and descriptive. Cognitive

methods, on which the AUS concept and construct are based, have roots in the psychological and

neurophysiological processes of understanding and describing analysed data as they take place in the human

brain.

1 INTRODUCTION

There are more and more intelligent information

systems being developed. Such systems, which was

evaluated as helpful in years of prosperity, are

indispensable in crisis time. Economical

competition, always very hard, become fierce in

contemporary situation. Therefore new methods of

IT-based management improvement tools are today

necessary more than whenever. The development

time of new information systems – decision support

systems is expected to become shorter and shorter

despite a growing complexity. Everybody looks for

new ideas, because only original methods of

successful management can to bear fruit in

competition superiority. Information systems

designed and developed for the future must have

new features and new roles. The are no more only

useful calculator, helping in routine management

procedures. The information system of the future

must be a weapon, revolutionising the way we

conduct business nowadays. Such revolutionary idea

of the use automatic understanding systems in

business practice is presented in this paper. This idea

is not mature enough to be introduced to the

economical practice nowadays, but it is different

from typically used decision support paradigms and

therefore can be found interesting.

Modern production, service, trading companies,

banks, insurance corporations, and the dispatch

industry, rely on information systems, which are

highly integrated in an organisation. However, they

do not take significant advantage of the most recent

advancements in artificial intelligence and cognitive

science. This is regretful, since the use of AI and

cognitive science could contribute significantly to IT

innovations and may constitute a source of

competitive advantage.

2 COGNITIVE ANALYSIS BASIC

METHODS AND IDEAS

More ambitious tasks, such as a strategic idea

generation, require a new approach. An AUS is a

natural way to go.

Tadeusiewicz R., Ogiela M. and Ogiela L. (2009).

AUTOMATIC INFORMATION PROCESSING AND UNDERSTANDING IN COGNITIVE BUSINESS SYSTEMS.

In Proceedings of the 11th International Conference on Enterprise Information Systems - Artiﬁcial Intelligence and Decision Support Systems, pages

5-10

DOI: 10.5220/0001844200050010

 SciTePress

However, it is worth noting that we are aiming to

develop an IS capable of using both the form and the

content of business data stored and retrieved for

analysis. To manage a business efficiently and, in

particular, to find and implement new concepts, it is

necessary to understand the situation of the

business as well as that of the business environment

(or of a given market segment). By understanding

the micro- and macroeconomic situation well, one

can have the basis to propose innovative changes,

which may even turn out to be revolutionary in

nature. However, the lack of understanding of

certain signals and/or data could lead to disaster.

From the viewpoint of psychological sciences, in

the process of understanding any information

obtained by a man, subject to cognitive analysis,

there are three stages:

 Registration, remembering and coding the

information obtained.

 Preserving – a latent stage of natural processes.

 Information reproduction – its scope covers the

remembering, recognition, understanding and

the learning of some skills anew (Ogiela,

Tadeusiewicz, 2003; Tadeusiewicz, Ogiela,

2004).

Some features of cognitive processes on which

we shall base our AUS information model, the

model analysed in this paper, can be associated with

the neural network technique, which is quite

fashionable nowadays. These networks are capable

of registering, remembering, storing and reproducing

various signals and other information. Nevertheless

the capacity of neural networks is limited by the fact

that in order to build them one uses rather small sets

of units - learning by processing information

(artificial neurons). As a result, within their scope lie

only very simplified models of psychological

processes - at best connected with the ability to learn

simple reflexes or elementary associations. The

neurophysiological model of cognitive analysis that

we need in this paper is based on the functioning and

operation of large brain fragments that can be

described by examining (among others) large neural

group dynamics attractors (amounting even to

millions of neurons), defined in the stimulation area

of these neurons (Tadeusiewicz, Ogiela, 2005).

Even though there are no identical states of brain

surface (for example examined with EEG), in the

dynamics of its activities one can find constant

relations between these attractors, i.e. relatively

repetitive dynamic neurophysiological states.

Appropriately, mathematically interpreted dynamic

states of the brain are characterised by some deeper

relations, which could have their logical

representation. The correspondence between the

state of mind and brain does not refer to the

electrophysiological surface phenomena of a volatile

nature, but it points to some stability of attractor

states.

Let us now try to shift these statements to a

cognitive IS model, of interest for us and used for

business purposes. Its task would be to interpret

facts based on the understanding and reasoning

conducted in connection with the semantic content

of processed data.

3 SEMANTIC ANALYSIS OF THE

DATA

Every Information System is supposed to perform a

semantic (directed at the meaning) analysis of

a selected object, or the basis of the information

must contain some knowledge necessary to make

a correct meaning analysis. Confronting the obtained

description of a currently analysed object (e.g. a

specified market situation), we obtain the premise to

show the real meaning and sense of the said object,

that is to understand it via its featured

characteristics along with its given set of

expectations and hypotheses. This is true for every

system capable of understanding any data and

information. This is due to the fact, that it is always

the knowledge held previously, i.e. the basis to

generate system expectations, that can constitute the

reference point for semantic analysis of features

obtained as a result of the conducted analysis of

every object, analysed at the system input. As a

result of the analysed objects combined features

together with the expectations generated, based on

the knowledge regarding its semantic content, we

find a cognitive resonance phenomenon (figure 1).

That is the key to meaning analysis of objects or

information.

Figure 1: The cognitive resonance phenomenon in the

process of the analysed object understanding.

ICEIS 2009 - International Conference on Enterprise Information Systems

4 NEW OBJECT

REPRESENTATION

In accordance with the concept developed by the

authors over a number of years, cognitive analysis

used in IS dedicated to automatic understanding, is

based on the syntactic approach. For the purposes of

meaning analysis and interpretation of the analysed

object it therefore uses a linguistic description

(Meystel, Albus, 2002; Tadeusiewicz, Ogiela, 2004).

As a result of the implementation of the pre-

processing stages described above as well as with

reference to mathematical linguistics, it is possible

to obtain a new object representation in the form of

hierarchic semantic tree structures and the

subsequent steps deriving this representation from

the start symbol of the grammar used. An intelligent,

cognitive IS, at the stage of pre-processing, must in

most cases decompose the analysed data to a lower

level of instance, identify primitive components and

determine the relationships between them. An

appropriate classification makes it possible to find

out whether the representation of a given object

belongs to a class of objects generated by a formal

language defined by a possible grammar defining

formal languages (i.e. sequential, tree or graph

languages). A recognition with their use takes place

in the course of syntactic and semantic analysis

performed by the system.

Further system operations relate to techniques

originating from mathematical linguistics. The

details of its use can be found in next chapters.

Nevertheless due to the fact that we are referring to

it for the first time here, it is worth explaining why

we decided to use this very tool in our papers. Well,

the basic difference one can see between automatic

data classification, frequently used in information

science, and the automatic understanding of the data

is that in the case of classification one knows in

advance a certain finite number of classes. The

recognition process is only supposed to determine to

which class one can include the object analysed at a

given time. On the other hand, if one tries to

understand the meaning of a specified object, we

have no “a priori” known list of possible meanings

to choose from. The sense of each data set must be

built though from scratch. One can say therefore that

the number of possible meanings detected in the

automatic understanding process is potentially

infinite; this is even true in one given situation when

we are dealing only with one determined meaning.

For this reason we need such an IT tool that would

be capable of generating an infinite number of

analysed object descriptors. Due to the

implementation possibility the tool itself must be

composed of a finite number of elements. Language

is such a tool. A natural language enables one to

express an infinite variety of moods and content

using a finite number of words. On the other hand,

artificial languages, for example algorithmic ones,

enable, in a similar way, the creation of an infinite

number of various software items. In our paper we

refer therefore to a formalised term of data

description language and to mathematical methods

and to the rules governing the processing of it. This

is to create the basis for the automatic generation of

(potentially) an infinite number of various meanings

using just a small set of elements, formal rules

(enabling computer application) and axioms; these

would form the grounds for an automatic

understanding system (Skomorowski, 2000).

5 AUS-TYPE INFORMATION

SYSTEM: A WHOLE CONCEPT

Figure 2: Division of functions between people and

computers in a traditional IS supporting business decision-

making.

We shall now present a proposed structure and

operational methods for the previously introduced

AUS-type system. First, in order to systematise our

considerations and to establish a reference point, let

us recall the traditional (nowadays applied in

practice) structure of an information system

application: computers are, obviously, involved

since they are the ones that store and process data as

well as analyse data in various ways. Information

obtained from such computer systems is entirely

sufficient for an effective management of business

processes at the tactical level (as marked jokingly on

AUTOMATIC INFORMATION PROCESSING AND UNDERSTANDING IN COGNITIVE BUSINESS SYSTEMS

Figure 3: General AUS-type information system structure.

figure 2). On the other hand, if we talk about

management at a strategic level, we find out that

despite automated data collection, storage and

analysis, the task of business meaning understanding

of the said data is in traditional systems the unique

area of people (experts). So is taking and

implementation of strategic decisions: this belongs

only to people holding appropriate, high positions.

The structure of such traditional IS, as presented on

Figure 2, will be the starting point in proposing a

general structure of an AUS-type information

system.

In such a system, whose structure is presented in

figure 3, the initial processes of storing and pre-

processing phenomena taking place in the analysed

business entity, are analogous to the one we are

dealing with in traditional systems. The only

difference is that with the perspective of automatic

interpretation of data analysis process results, one

can compute and collect a larger number of ratios

and parameters since the interpreting automaton will

not be dazzled or perplexed by an excess of

information. This is what happens when people,

interpreting situations, are ‘bombed’ with hundreds

of ratios among which they can hardly find the

important ones and then need to make a huge effort

in order to interpret them correctly. There may also

be no change to the business process management at

the tactical level. This was left out of figure 3

entirely since the AUS information systems concept

does not refer to this level at all.

The difference between a traditional system and

the AUS becomes visible when the computer

independently, and without human participation,

attempts to describe the properties and consequences

of the ratios computed. The results of automatic

interpretation are expressed in the categories of the

applied description language for the interpreted data

properties. The above-mentioned language is a key

element at this stage. It must be designed with great

know-how. Its construction must therefore be based

on collecting and systematising expert knowledge.

Referring to the analogy with medical image

automatic understanding systems, which were earlier

said to be the area of some fully successful

implementations of ideas described here, one could

say that just like in medical systems (Ogiela,

Tadeusiewicz, 2003) the basis for the development

of the language subsequently used for semantic

image interpretation (and diagnosing a disease) were

some specified changes in the shape of the analysed

organs. In the AUS-type information systems the

basic constructing units of the developed language

should be changes of some specified business

indicators.

6 APPLICATION OF

MATHEMATICAL

LINGUISTIC APPROACH AS A

KEY POINT OF NEW IDEA

Of course, focusing attention on a business index

and ratio changes computed in the input part of the

analysed AUS-type information system, as only on

those elements, which should be the basic

components of an artificial language, is just the first

step. The listing and appropriate categorisation of

ICEIS 2009 - International Conference on Enterprise Information Systems

changes that should be registered in linguistic

business processes corresponded only to the stage at

which one defines the alphabet to be later used to

build words and sentences, i.e. the language main

object. In order to make it possible to create from the

elements of this alphabet counterparts of words and

sentences for subsequent use by the AUS to describe

the states of business process, which require

understanding and interpretation, calls for an

introduction of additional mechanisms. These

mechanisms would enable combining the above-

mentioned sentences into larger units. Therefore, at a

level superior to the above-described alphabet one

must build the entire grammar rules and

transformations. This grammar can then be used to

create complete languages of description expressing

important content, necessary to understand

automatically the analysed processes.

In medical image understanding systems we

constantly refer to, at our disposal were tools

detecting local changes in the shape of some

specified internal organs and their morphologic

structures (Tadeusiewicz, Ogiela, 2004). These were

the above-mentioned ABC-base. To understand the

state of a given organ correctly, one needed to add to

these graphic primitives their mutual spatial

relations and combine them with anatomy elements.

Owing to a definition of rules and the grammar

constructs connected with them, one could combine

for example a graphic category ‘change of edge line

direction of a specified contour’ with a meaning

category ‘artery stenosis anticipating a heart failure.’

Similarly, by building into the proposed

language grammar the ability to associate business

changes detected in various parts of the managed

company and its environment as well as the

possibility to trace and interpret time sequences of

these changes and their correlations, it will be

possible, for example, to understand what are the

real reasons behind poorer sales of goods or services

offered. As a result it will be, for example, possible

to find out about the fact that this is due to the wrong

human resources policy rather than the wrong

remuneration (bonus) system.

After the development of an appropriate

language which will (automatically!) express

semantically oriented descriptions of phenomena

and business processes detected in the business unit

(e.g. a company) as supervised by the information

system, a further AUS information system operation

will be very similar to the structure in which

function the medical systems previously built by this

system authors. The starting point for the business

data automatic interpretation process, the process

finishing with understanding their business meaning,

is the description of the current state proposed in the

system. It is expressed as a sentence in this artificial

language, built specially for this purpose. Without

going into details (described, among others, in

earlier publications listed in the bibliography) one

can say that the above-mentioned language

description for a human being is completely illegible

and utterly useless. A typical form of such notation

is composed of a chain of automatically generated

terminal symbols. Their meaning is well based in the

mathematically expressed grammar of the language

used. Yet from the human point of view this notation

is completely illegible.

7 CONCLUSIONS

This paper presented the problem of divergence of

IS to accommodate needs imposed by current

changes to business. The disparity between the

needs and real capacity of IS can be observed by

analysing the computer management support

association with a need for making strategic

decisions. To make this term clear let us specify that

in this paper, we understand that all decisions taken

at various levels of company management can be

perceived as strategic if they are not limited to a

simple regulation of stable business processes but

which induce and impose changes. Therefore,

strategic decision making cannot be made solely on

the basis of information about the current state of

affairs in the on-going business processes. Their

essence can change the state of affairs. For this

reason, current IS, focused mainly on registration

and settlement functions are insufficient to support

such decision-making processes.

The analysed steering functions, as has been

realised, are decisions that go far beyond simple

tactic management, and are executed by issuing

detailed commands and giving simple tasks whose

execution can be assessed based on a small number

of well-defined parameters. There are a number of

Information Systems that support tactic decisions

but for a real support of strategic decisions there are

practically no acceptable methods. At the same time,

the growing complexity of modern business

processes, taking place in the conditions of global

economy, results in that more and more people have

to make such difficult decisions. Many of the

decision makers are not ready prepared for this

situation, neither professionally nor mentally.

It makes sense to support their work with IS of

the proposed AUS-type information system. The

AUTOMATIC INFORMATION PROCESSING AND UNDERSTANDING IN COGNITIVE BUSINESS SYSTEMS

system would be equipped with previously

unavailable methods to process of the data semantics

for supporting the management in the strategic

decision making.

We have outline the concept structure that could

be the basis for such an AUS-type system dedicated

to the automatic business data understanding. We

have tried to demonstrate briefly that a will to build

such a system is an objective worth aiming at.

In the most competitive global market, the

economy cannot be treated as a “zero-sum” game in

which the success of some businesses must

necessarily be based on (or depends on) the failure

of others.

The economy is not a zero-sum game, which

suggests that the success of one particular business

does not infer the automatic failure of another

business. Quite the contrary, the whole global

economy is more and more oriented towards looking

for solutions that can be referred to as “win-win

solutions”, i.e. solutions resulting in success for all

participants, although each of them may be active in

different fields that have been achieved via different

scopes. For such an economy, implementing the

innovation based on the AUS concept and on

cognitive premises and methods, this should be

interpreted not as a source of threat but rather as

another factor for global growth and development.

ACKNOWLEDGEMENTS

This work has been supported by the AGH

University of Science and Technology under Grant

No. 10.10.120.783

REFERENCES

Meystel, A.M., Albus, J.S., 2002. Intelligent Systems –

Architecture, Design, and Control. A Wiley-

Interscience Publication John Wiley & Sons Inc.

Ogiela, L., Tadeusiewicz, R., Ogiela M.R., 2008.

Cognitive Categorizing in UBIAS Intelligent Medical

Information Systems, in Sordo M., Vaidya S., Jain

L.C. (eds.): Advanced Computational Intelligence

Paradigms in Healthcare 3, Studies in Computational

Intelligence 107, Springer-Verlag, Berlin, Heidelberg,

2008, pp. 75-94.

Ogiela, L., Tadeusiewicz, R., Ogiela, M.R., 2007.

Cognitive Categorization in Modeling Decision and

Pattern Understanding. In: Torra V., Narukawa Y.,

Yoshida Y.: Modeling Decisions for Artificial

Intelligence, MDAI 2007 CD-ROM Proceedings,

ISBN 978-84-00-08539-1, pp. 69-75.

Ogiela, M.R., Tadeusiewicz, R., 2003. Artificial

Intelligence Structural Imaging Techniques in Visual

Pattern Analysis and Medical Data Understanding.

Pattern Recognition (pp.2441-2452). Elsevier vol.

36/10.

Skomorowski, M., 2000. A Syntactic-statistical approach

to recognition of distorted patterns. UJ. Kraków.

Tadeusiewicz, R., Ogiela L., Ogiela M.R., The automatic

understanding approach to systems analysis and

design, Elsevier, International Journal of Information

Management 28 (2008) pp. 38-48.

Tadeusiewicz, R., Ogiela, L., 2008. Selected Cognitive

Categorization Systems, Chapter in book: Rutkowski

L. et al. (Eds.): Artificial Intelligence and Soft

Computing, ICAISC 2008, Lecture Notes on Artificial

Intelligence, vol. 5097, Springer-Verlag Berlin

Heidelberg, pp. 1127–1136.

Tadeusiewicz, R., Ogiela, L., 2008. Modern Methods for

the Cognitive Analysis of Economic Data and Text

Documents and Their Applications in Enterprise

Management. In: Snasel V., Abraham A., Saeed K.,

Pokorny J. (eds.) Proceedings 7th International

Conference on Computer Information Systems and

Industrial Management Applications CISIM 2008,

IEEE Computer Society, IEEE, Los Alamitos,

California, pp. 11 – 23.

Tadeusiewicz, R., Ogiela, M.R., 2004. Medical Image

Understanding Technology. Springer-Verlag Berling

Heildelberg.

Tadeusiewicz, R., Ogiela, M.R., 2005. Intelligent

Recognition in Medical Pattern Understanding and

Cognitive Analysis. Chapter in book Muhammad

Sarfraz (Eds.). Computer-Aided Intelligent

Recognition Techniques and Applications. (pp. 257-

274). John Wiley & Sons Ltd.

Zadeh L.A., 2008. Toward human level machine

intelligence--Is it achievable? Proc. 7th International

Conference on Cognitive Informatics (ICCI’08), IEEE

CS Press, Stanford University, CA.

Zhong N., Raś Z.W., Tsumoto S., Suzuki E. (eds.), 2003.

Foundations of Intelligent Systems, 14th International

Symposium, ISMIS 2003, Maebashi City, Japan.

ICEIS 2009 - International Conference on Enterprise Information Systems