=Paper=
{{Paper
|id=Vol-2830/paper29
|storemode=property
|title=Cognitive Modeling of University Activity as a Poorly Structured System
|pdfUrl=https://ceur-ws.org/Vol-2830/paper29.pdf
|volume=Vol-2830
|authors=Andrey Mikryukov,Mikhail Mazurov,Dmitry Korneev,Vasilii Trembach,Davyd Rylenkov
}}
==Cognitive Modeling of University Activity as a Poorly Structured System==
https://ceur-ws.org/Vol-2830/paper29.pdf
Cognitive Modeling of University Activity as a Poorly
Structured System
Andrey Mikryukov 1[0000-0002-8206-677X], Mikhail Mazurov 1[0000-0001-9993-4687],
Dmitry Korneev 1[0000-0001-7260-4768], Vasilii Trembach 2[0000-0001-7499-4368],
Davyd Rylenkov 1[0000-0002-0198-4246]
1
Plekhanov Russian University of Economics, 36 Stremyanny lane, Moscow, 115998, Russia
Mikrukov.aa@rea.ru, Mazurov37@mail.ru,
Korneev.dg@rea.ru, Dave.Rylenkov@gmail.com
2
Moscow Aviation Institute (National Research University),
4, Volokolamskoe shosse, Moscow, 125993, Russia
Trembach@yandex.ru
Abstract. The relevance of the problem being solved is due to the need to
develop scientifically grounded proposals to achieve the required values of the
basic indicators of the university's activities in accordance with the international
institutional rating QS to the required values necessary for the university to
enter the TOP-500 universities by 2025. To solve this problem, an approach is
proposed to the study of weakly structured systems, the class of which includes
universities and their activities, based on scenario forecasting methods by
building a fuzzy cognitive model in order to determine the necessary increments
of target values. The proposed approach allows, under the given constraints, to
find the most acceptable scenario for planning the increment of basic indicators
to target values by identifying the latent factors influencing them and impulse
influences (increments) on them, ensuring guaranteed achievement of the set
goal. The results obtained make it possible to subsequently justify the annual
costs to ensure an unconditional increase in the values of latent factors (private
indicators) to guarantee the required values of the basic indicators by 2025. The
novelty of the proposed approach lies in the use of correlations between latent
factors, identified on the basis of factor analysis methods, with basic indicators
in the construction of a fuzzy cognitive model, as well as the application of an
iterative approach to solving the problem, which makes it possible to update
every year according to the results of the next ranking of universities in the
international institutional ranking a set of initial data, as well as train the
developed cognitive model, taking into account the results of identifying latent
factors and adjusting their correlations. The results obtained make it possible to
form the most preferable scenario plan for the necessary stepwise increase in
the values of basic indicators in the interval 2020 -2025 subject to resource
constraints.
Keywords: Cognitive model, scenario forecasting, baseline indicators,
institutional ranking.
Proceedings of the 10th International Scientific and Practical Conference named after A. I. Kitov
"Information Technologies and Mathematical Methods in Economics and Management
(IT&MM-2020)", October 15-16, 2020, Moscow, Russia
© 2021 Copyright for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org)
1 Introduction
The purpose of developing a cognitive model of scenario forecasting is to
substantiate the list of necessary measures to ensure the increments of the values of
the target indicators of the university's activities in accordance with the international
institutional rating QS to the required values necessary for the university to enter the
TOP-500 universities by 2025.
To achieve the stated goal of the study, the application of methods for solving
weakly structured tasks was substantiated based on the development of a scenario
forecasting (planning) model using fuzzy cognitive maps, which made it possible to
determine probable (possible) trends in the development of events by alternative
options and the possible consequences of decisions made in order to choose the most
preferable one. alternatives. The proposed approach allows, under the given
constraints, to find the most acceptable scenario for planning the increment of the
values of basic indicators and functional (university rating) to target values due to
impulse influences on latent factors (assessing the feasibility of the necessary
increments in the values of factors - causes), ensuring guaranteed achievement of the
goal.
It should be borne in mind that the value of the functional and the place in the QS
rating from year to year are not one-to-one, since the number of universities
participating in the QS rating changes annually towards their increase. It follows from
this that the value of the functional corresponding to a certain place in the rating in the
current year may not correspond to the same place in the rating in subsequent years.
That is, the value of the functional must be calculated with a certain margin.
In the course of the study, the following tasks were solved: a fuzzy cognitive
model of scenario forecasting of measures to achieve the required values of the target
performance indicators of the university in the international institutional ranking QS
was developed, on the basis of the developed model, the calculation of the most
preferable variant of the set of required intensities of influence on control variables
(factors - causes) was carried out on the basis of the developed model. a given
increment of the value of the target factor (functional). The scenario of planning
measures for guaranteed achievement of the target value obtained on the basis of the
cognitive model made it possible to substantiate the value of the necessary increments
in the values of the identified latent factors that affect the baseline indicators. Based
on the results obtained, in the future, it is possible to justify the necessary annual costs
to ensure the guaranteed achievement of the required value of the rating functionality
of the university by 2025.
The results obtained made it possible to formulate a scenario plan for the necessary
stepwise increase in the values of indicators, considering the factors influencing them.
A feature of the developed approach is the identification of the most significant
latent factors affecting the baseline indicators, as well as the assessment of the degree
of their influence on the baseline indicators based on the methods of factor analysis.
To ensure the adequacy of the cognitive model, an iterative approach was used, which
makes it possible to build a learning cognitive model. The use of this approach
provides for the correction of the cognitive model based on the results of assessing the
activities of universities after the release of the annual report on the place of the
university in the QS institutional ranking. The adjustment is carried out by clarifying
the structure of the cognitive map and causal dependencies based on updating the sets
of initial data, as well as the results of identifying latent factors using a factor model.
The use of an iterative approach allows for an increase in the reliability and
accuracy of obtaining a solution.
Section 2 provides a literature review on the research topic, Section 3 presents the
results of the development of a cognitive model, Section 4 presents the results of
scenario forecasting of university performance indicators based on the developed
cognitive model of scenario forecasting, Section 5 provides an interpretation of the
results of a numerical experiment.
2 Literature Review
A fairly large number of works by domestic and foreign scientists are devoted to
the problem of cognitive modeling [1-14]. An important task when using cognitive
models is to ensure their reliability, adequacy, and accuracy. In papers [1-4], an
approach based on classical cognitive models is proposed. In work [5], problem
solving using fuzzy cognitive models is considered. The work [7] solves the problem
of assessing the stability of cognitive models, as well as methods for their verification.
In works [8,11-14], the features of modeling nonlinear dynamic systems under
uncertainty are considered.
The analysis of the sources showed the expediency of using the class of fuzzy
cognitive models and that in the presented formulation, the problem of cognitive
modeling of university development scenarios was not considered.
The choice of a fuzzy cognitive model is based on the following reasons. Classical
cognitive maps do not always provide for the construction of an adequate and reliable
mathematical model due to the high level of uncertainty in the interaction of the
components of the research object. They are linear and do not fully consider the
dynamic properties of real systems, which can be essentially nonlinear and
nonstationary. The strength of connections between concepts of a fuzzy cognitive
map is set using a fuzzy linguistic scale, which is an ordered set of linguistic
meanings (terms) of estimates of the strength of connection. The use of fuzzy
cognitive maps provides a convenient interpretation of causal relationships between
concepts. All this taken together ensures the construction of a more adequate model of
the object of research.
3 Solving Problem Methodology
The cognitive modeling methodology is an independent research area that is
actively developing. Cognitive maps are a convenient mathematical apparatus for
describing and studying poorly structured systems (socio-economic, socio-technical,
organizational, etc. systems), the distinctive features of which are functioning under
conditions of uncertainty, characterized by a lack of information, processes occurring
in them, the impossibility of analytical descriptions and construction of formal models
that take into account the specifics of the systems under consideration.
The analysis of the problem posed showed that it belongs to the class of semi-
structured problems, which can be solved under the conditions of a limited amount of
initial data and a number of uncertainties associated, among other things, with the
absence of a linear relationship between the values of the functional and the place of
the university in the QS ranking in the considered time interval.
To solve this class of problems, the methodology of fuzzy cognitive modeling is
used, designed for analysis and decision-making in poorly defined situations when it
is not possible to analytically describe and build formal mathematical models that
take into account the specifics of the studied poorly structured system and, in
particular, the method of scenario forecasting (planning ) based on cognitive maps
(cognition maps), which allows one to determine the probable (possible) trends in the
development of events by alternative options and the possible consequences of
decisions made in order to choose the most preferable alternative [14].
A cognitive model in the form of a fuzzy cognitive map is a subjective model of a
weakly structured dynamically developing situation.
As you know, a scenario is a certain relative, conditional assessment of the
possible development of the object or situation under study, since it is always built
within the framework of assumptions about future development conditions, which are
often fundamentally unpredictable.
In contrast to classical forecasting methods, in which the main attention is paid to
assessing the most probable variant of system development, scenario forecasting
proceeds from the idea of uncertainty and ambiguity of the trajectory of this
development. A feature of scenario forecasting is that it allows you to simultaneously
consider several options for the development of the situation, analyze opportunities
and risks [12].
In the context of the problem under consideration, a scenario is understood as a
dynamic sequence of possible events of change in the values of factors - causes and
factors - consequences that affect the target indicators. Causal relationships between
these events and decision points can change their course and the trajectory of
movement in time for the entire system of indicators under consideration, and,
therefore, choose the most preferred trajectory option.
To solve this problem, an approach to scenario forecasting based on a fuzzy
cognitive map is proposed, which includes the following stages [14]:
- scenario generation and risk assessment,
- scenario adjustment at discrete time intervals (in our case, annually), considering
the achieved values of the place in the QS rating of the functional and indicators.
At the end of the next time interval, a new scenario is built to achieve the next set
goal (adjusted considering current calculations).
At the first stage of the study, the identification and interpretation of the latent
factors affecting the baseline indicators was carried out, and their significance was
assessed using the methods of factor analysis, which is a class of multivariate
statistical analysis procedures aimed at identifying the latent variables (factors)
responsible for the presence of linear statistical relationships. (correlations) between
the observed variables [15-18].
Factors are groups of certain variables that correlate with each other more than
with the variables included in another factor. Thus, the meaningful meaning of the
factors was revealed by examining the correlation matrix of the initial data.
In order to identify the factors influencing the basic indicators of the university's
activity, to identify the most significant factors for the subsequent implementation of
the approach to solving the problem on the basis of cognitive modeling methods, their
SWOT analysis was carried out [19-21].
As you know, the SWOT analysis method involves a deep analysis of the research
object, provides the most objective assessment of it from the point of view of the
strengths (positive) and weak (negative) sides of the external and internal
environment, as well as opportunities and threats. The results of the SWOT analysis
made it possible to build the problem field of the situation, on the basis of which the
goals and objectives of cognitive modeling were formulated, as well as the structure
of the cognitive map was determined, with the help of which the task of predicting the
performance indicators of the university was solved. When constructing the problem
field of the situation for the structuring of knowledge, an object-structural approach
was used [9], according to which the analysis and presentation of knowledge is
carried out in strategic, organizational, conceptual, functional, spatial, temporal,
causal and economic aspects (strata).
The results of the analysis made it possible to structure the knowledge of experts
using the problem field of knowledge, to build and identify on the basis of the method
of expert assessments a set of factors and the degree of their influence on the
performance indicators of the university.
Registration of SWOT- analysis data was carried out in tabular form, where the
main elements were recorded according to the categories presented (Table 1).
Table 1. SWOT analysis results.
The analyzed factors (characteristics) The degree of The degree of
embodiment of the factor importance of the
(characteristics) factor
Strong factor + (characteristics)
Weak factor –
1. Availability of well-known scientific Strong factor + 0,6
schools and dissertation councils
2. The presence of close collaboration with Strong factor + 0,3
foreign universities and research organizations
(the number of joint research projects
3. Availability of basic departments at Strong factor + 0,2
enterprises
… …. …..
Factor (characteristic) N …. ….
The result of the SWOT analysis was the identification and grouping of a set of
factors that affect the performance of the university. Since the number of factors
influencing the activities of the university is a significant value, it became necessary
to highlight the most significant factors, taking into account the correlation
relationships, including the factors of the second level that affect the factors of the
first level.
In order to identify the most significant factors, one of the most common methods
of factor analysis was used - the method of principal components, which allows one to
reduce (reduce) a large number of related correlated variables, since a large number of
variables significantly complicates the analysis and interpretation of information [17].
Calculations based on the method of principal components were carried out in the
environment of the analytical platform Deductor 5.3 using the "Factor analysis"
module.
The use of the apparatus of factor analysis made it possible to solve two main
problems:
1) reducing the dimension of the number of variables used due to their explanation
by a smaller number of factors;
2) grouping and structuring of the received data.
The novelty of the proposed approach is determined by:
- solving the problem of structuring knowledge using the object-structural
approach when constructing the problem field of the situation, which provides a more
effective grouping of latent factors and building an adequate architecture of the
cognitive map;
identifying the most significant latent factors affecting the baseline indicators, as
well as assessing the degree of this influence based on the methods of factor analysis;
- the implementation of a cognitive model based on a fuzzy cognitive map, which
provides adaptation to the uncertainty of the initial data and the conditions of the
problem being solved, which, in turn, made it possible to form a more adequate
cognitive model and increase the reliability and accuracy of modeling results;
- using an iterative approach to building a cognitive model that allows you to build
a trained cognitive model;
- the possibility of adjusting the fuzzy cognitive model based on the results of the
annual assessment of universities' performance in the QS institutional ranking. The
adjustment is carried out by clarifying the structure of the cognitive map and causal
dependencies based on updating the sets of initial data, as well as the results of
identifying latent factors using the factor model.
3 University Performance Indicators Scenario Forecasting Results
Based on the Cognitive Model
In contrast to classical cognitive maps, fuzzy cognitive maps (FCM) are represented
as a fuzzy directed graph with feedback, the nodes of which are fuzzy sets. FCM
combines the properties of fuzzy systems and neural networks [14].
As a result of the selection of the most significant factors affecting the baseline
indicators, a cognitive map was built that reflects their interconnections, as well as the
weight of these interconnections (Fig. 1). The weights of the relationships are
determined based on the use of correlation analysis and expert methods.
Fig. 1. Cognitive map of the scorecard based on the graph of the relationship of factors,
baseline indicators and functionality (Legend: F - functional; R - rating, (F1-F24) - latent
factors)
The following designations are adopted in Figure 1: F is functional; R is university
rating, AR is academic reputation; EP is reputation with the employer; RS/T is the
ratio of the number of students to the number of teachers; CT - indicator of citation of
teachers; IT is number of international teachers; IS is the number of international
students; F1 is factor "Scientific schools and dissertation councils" (0.6); F2 is factor
"Joint research projects" (0.3); F3 is Presence of basic departments (0.2); F4 is
Number of publications in the Scopus database (0.6); F5 is In-demand areas of study
(0.3); F6 is skill level of the teaching staff (0.2); F7 is Number of teachers (0.6); F8
is Students' Competence Level (0.5); F9 is teachers with language training (0.4); F10
is Dorm Places (0.2); F11 is Demand for graduates from employers (0.3); F12 is
Areas for educational activities (0.3); F13 is The level of payment for teaching staff
(0.4); F14 is Stimulating factors (0.2); F15 is Expansion of the CPD social package
(0.3); F16 is Change in the structure of employment of the CPD (0.3); F17 is
Proportion of teaching staff planning to pursue an international scientific career (0.2);
F18 is Academic mobility of the teaching staff (0.3); F19 is Convergence of
educational programs with foreign universities (0.4); F20, is Foreign Entry Company
(0.3); F21 is Increase in the number of On-line MOOCs (0.3), F22 is Implementation
of individual educational trajectories (0.4); F23 is Implementation of remote sensing
technologies (0.3); F24 is The tightness of the relationship with the employer (0.4).
The directed graph (digraph) of the FCM is defined using a set of sets [14, 22-28]
FCM = <С, F, W>,……… ……..…..………… (1)
where C = {Ci} is the set of graph vertices, called concepts, which are factors that are
most significant from the point of view of studying the system under consideration;
F = {Fk} - set of directed graph arcs - connections between concepts;
W = {Wij} is the set of weights of arcs (links).
The connections between concepts can be positive, “enhancing” the influence of
the Ci concept on the Cj concept (Wij> 0), or negative, “weakening” the influence of
the Ci concept on the Cj concept (Wij <0).
Each concept is characterized by a term-set of a linguistic variable
Ti = {T1i, T2i ,…, Tmji} (2)
where mj is the number of typical states of the i-th concept. To describe each term
Tki, a term is constructed - a set with a membership function . The connections
between the typical states of each pair of concepts are set by fuzzy variables described
by the corresponding fuzzy sets.
The values of the weights (bond strength) Wij are set using a fuzzy linguistic scale,
which is an ordered set of linguistic values (terms) of bond strength estimates, for
example, of the form: LINK_POWER = {Does not affect; Weak; Average; Strong;
Very strong}. Each of these values is associated with a certain numerical range
belonging to the segment [0, 1] for positive links (Table 2), or the segment [–1, 0] for
negative links.
Table 2. Assessment of the strength of the connection between concepts.
Linguistic value Term designation Numerical range Point estimate of
bond strength +
0 Z 0
Does not affect
(0; 0,15) VL 0,12
Very weak
(0,15; 0,35) L 0,23
Weak
(0,35; 0.60) M 0,47
Average
(0,6; 0,85) H 0,72
Strong
(0,85-1,0) VH 0,93
Very strong
The strength of the connection between the concept Ci and the concept Cj is chosen
by the expert by one of the linguistic values presented in the table, as well as by some
"point" estimate of the strength of the connection - a number within this range (if
there are several experts, then the weight Wij is averaged).
In the general case, a weighted digraph with arbitrary values of the weights Wij
–1, 1 is described by the dynamics of its state change in time. The state of the
digraph (FCM) is determined by the set of states of its concepts Ci , (i = 1, 2, …, n),
each of which is described by the state variable Xi (t), which takes values in the
interval [0, 1].
State variables Xi (values of latent factors) are calculated by normalizing real
("physical") variables by the formula
Xi = , (3)
where Хi min and Хi max are the minimum and maximum values of the variable Хi, (i = 1,
2,…, n).
The state of the FCM at an arbitrary moment of time is described by the equation
of state of the following form [29]:
The weighted digraph in Fig. 1 with arbitrarily set values of the weights Wij is
described by the equations of state in the following form:
where is some nonlinear “contracting” function that maps the values of the
argument to the unit interval [0, 1].
This condition is satisfied, for example, by the sigmoid function
(x) = (5)
To calculate the state variables Xi(t), (i = 1, 2, …, n) using equations (4), it is
necessary to set the initial conditions, i.e. vector X(0) = (X1(0), X2(0),…, Xn(0))T.
An important stage in the FCM analysis is the analysis of its impulse stability [25],
when for a given nonzero initial state Xi (0), for example, X1(0) = 1, X2(0) = …=
Xn(0) = 0, the sequence of impulse values pi (t) = Xi(t)- Xi(t-1) is bounded at any time
instant t = 1,2, ..., m for any of its vertices, and absolute stability, when for each
vertex of the digraph (i = 1, 2 ,…, n) the sequence of values значений Xi(t), t = 1, 2,
…m is bounded.
In this case, the statement [30,31] is true: the weighted digraph, which is described
by equations (4), (5), is absolutely stable, and there is a unique equilibrium (steady-
state) solution of the indicated equations ("fixed point") X* in that and only in that
case if
where n is the number of FCM concepts. Verification of the developed FCM using
formula (6) confirmed its stability.
The results of assessing the influence of latent factors on the baseline and rating
indicators led to the following conclusions. The least influence is exerted by the
factors: F3, The presence of basic departments at enterprises; F14, Stimulating
factors. In turn, the following factors have the greatest influence: F1, The presence of
well-known scientific schools and dissertation councils; F2, Close collaboration with
foreign universities and research organizations (number of joint research projects; F4,
Number of publications in the Scopus database, WoS.
Based on setting the values of the initial data using formula (3), setting the weights
of the mutual influences of factors using expert estimates and the results of
correlation-regression analysis, describing the behavior of the digraph using equations
of state (4), (5) and indicating the values of the initial increments of factors, analysis
of the dynamics of changes in factors and their influence on the development of the
system of indicators as a whole.
Verification of the developed cognitive model to check its adequacy was carried
out by testing in the retrospective period 2014-2019. based on available statistics on
measurable factors in the model. The general correctness of the model at this stage
was confirmed by the closeness of the growth rates of factors calculated on the model
to the actual growth rates.
The control problem is considered solved if, under the conditions of the given
constraints, an alternative option (planning scenario) is found to increment the values
of the functional and indicators to the target values due to impulse influences
(increments of factor values), which ensures the guaranteed achievement of the set
goal [32-37].
The “weak” increments of the values of latent factors at the level of 10%,
consistently set in the course of simulation modeling of controlled development of the
system of indicators, made it possible to evaluate the sensitivity of the target indicator
(rating functional) to control actions in these areas of regulation (Table 3).
Table 3. Results of assessing the sensitivity of the target to changes in latent factors.
Experiment Controlled Initial value Impulse of Change of rating
number factor change functional
1 F1 0,59 0,01 0,09
2 F2 0,19 0,02 0,07
… … … … …
24 F24 0,32 0,01 0,03
Considering the preliminary results of the computational experiment in the course of
scenario modeling, the required intensity of influence on the control factors in
percentage terms was calculated for a given increment of the target factor. The
problem was solved under the conditions of the given restrictions on the resources
allocated to ensure the increment of latent factors.
4 Discussion. The Results of a Numerical Experiment
Interpretation
Considering the achievable values of latent factors in the interval 2020-2021. the
scenario of their increment is selected, shown in Fig. 3. The greatest intensity of
impact will be required for latent factors: Number of publications in the Scopus
database, WoS, Joint research projects, Demand for areas of training, stimulating
factors, the least for latent factors: Changes in the structure of teaching staff
employment, Areas for educational activities, The share of teaching staff planning to
build an international scientific career.
The results of calculating the predicted values of growth indicators of basic
indicators and functional in 2020, considering the chosen scenario, are presented in
Fig 3. The obtained results of scenario forecasting show that the largest increase
should be obtained by the PP indicator (reputation with the employer) - 54%, and the
smallest - MP ( the ratio of the number of students to the number of teachers) - 16%.
An increase in the values of basic indicators should lead to an increase in the value of
the functional by 23%, i.e. achieving its planned value. Thus, the goal of the study has
been achieved.
Fig. 2. Calculated values of the control actions intensities on latent factors in the form of
increments in%, necessary to achieve the target increase in baseline indicators and functionality
in 2020
The results of calculating the predicted values of growth indicators of basic indicators
and functional in 2020, considering the chosen scenario, are presented in Fig 3. The
obtained results of scenario forecasting show that the largest increase should be
obtained by the PP indicator (reputation with the employer) - 54%, and the smallest -
MP ( the ratio of the number of students to the number of teachers) - 16%. An
increase in the values of basic indicators should lead to an increase in the value of the
functional by 23%, i.e. achieving its planned value. Thus, the goal of the study has
been achieved.
Scenario analysis of the development of the situation made it possible to choose
the most preferable scenario, which, under the conditions of the given restrictions,
ensures the achievement of the required planned value of the target indicator with
minimal resource consumption for the increment of latent factors.
The required intensity of influence on the control factors as a percentage for a
given increment of the target factor were calculated during scenario modeling. The
simulation results presented in Fig. 2, considering the attainable values of latent
factors in the interval 2019-2020. correspond to the scenario of their increment,
ensuring the achievement of the required target indicator (functional) with minimal
resource costs for the increment of latent factors.
Fig. 3. Forecasted values of basic indicators and functional growth in 2021. compared to 2020
(2020 -100%)
As a result of the calculations performed, the following conclusions can be drawn:
To guarantee the achievement of the functional value F = 26.89 by 2025,
corresponding to the 435 (451) place in the QS rating, it is necessary to stepwise
increase the values of indicators, taking into account the factors influencing them,
identified on the basis of factor analysis methods, in the interval 2020 -2025.
according to the following scheme.
Based on the calculation performed using the cognitive modeling methodology, it
is necessary to perform impulse influences (to provide an increment) on the set of
factors identified as of the present in accordance with the results presented in Fig. 2.
At the end of the control interval (after 1 year), after clarifying the actual values of
the basic indicators and functional in the QS rating, assess the change in the influence
and interaction of factors on the basic indicators based on factor analysis methods,
identify additional latent factors, update the developed scenario forecasting model
based on fuzzy cognitive maps taking into account additional latent factors, allowing
to build a more adequate and reliable model of the cognitive map. Based on the
updated model, perform calculations to justify a set of new impulse influences on the
latent factors of the cognitive model for the period 2021-2022.
5 Conclusion
During the research, an iterative learning cognitive model of scenario forecasting
of activities to achieve the required values of the target performance indicators of the
university in the QS international institutional ranking was developed. On the basis of
the developed model, the calculation of the most preferable from the point of view of
cost minimization of the scenario variant of the set of required intensities of influence
on the control variables (latent factors) for a given planned increment of the value of
the target factor was carried out.
The novelty of the proposed approach is determined using factor analysis methods
to determine the relationships between latent factors, which provide identification of
the most significant latent factors for building an adequate cognitive model. The
proposed approach allows training the developed model by adjusting the structure of
the cognitive map and its parameters based on the results of updating the initial data
for its construction.
5. Acknowledgments
Work is performed with assistance of the Russian Federal Property Fund, the
projects No. 18-07-00918, 19-07-01137, 20-07-00926.
References
1. Maksimov V. I, Kachaev S. V. Technologies of the Information Society in Action:
Application of Cognitive Methods in Business Management,
http://www.rfbr.ru/default.asp?doc_id=5222.
2. Axelrod R. Decision structure: cognitive maps of political elites. - Princeton. University
Press, 1976.
3. Kuznetsov O. P. Cognitive modeling of semi-structured situations //
http://posp.raai.org/data/posp2005/Kuznetsov/kuznetsov.html.
4. Roberts FS Discrete mathematical models with applications to social, biological and
environmental problems. - M .: Nauka, 1986 .-- 184 p. 13. Kosko B. Fuzzy cognitive
maps. // International Journal of Human and Machine Research, (1986) 24, 65-75.
5. Carvalho J.P., Tom J.A.B. Rule-Based Fuzzy Cognitive Maps - Fuzzy Causal
Relationships. // Computational Intelligence for Modeling, Control and Automation:
Evolutionary Computing and Fuzzy Logic for Intelligent Control, Knowledge and
Information Retrieval, edited by M. Mohammadyan, IOS Press, 1999.
6. Kulba VV, Mironov PB, Nazaretov VM Modern approaches to modeling complex socio-
economic systems of socio-economic systems using sign digraphs // Automation and
telemechanics - 1993. - №7.
7. Nikolay Karabutov, “Structural identifiability of nonlinear dynamical systems under
uncertainty”, International Journal of Intelligent Systems and Applications (IJISA),
Volume 12, No. 1, pp. 12-22, 2020. DOI: 10.5815 / ijisa.2020.01 .02.
8. Evgeny Bodyansky, Alexey K. Tishchenko, Anastasia O. Deineko, “An evolving neuro-
fuzzy system with online learning / self-learning”, IJMECS, volume 7, No. 2, pp. 1-7,
2014. DOI: 10.5815 / ijmecs .2015.02.01.
9. V. Vijayalakshmi, K. Venkatachalapati, Comparison of Predicting Student Performance
Using Machine Learning Algorithms, International Journal of Intelligent Systems and
Applications (IJISA), Volume 11, No. 12, pp. 34-45, 2019. DOI: 10.5815 / ijisa
.2019.12.04.
10. Gorelova G.V., Zakharova E.N., Rodchenko S.A. Study of semi-structured problems of
socio-economic systems: a cognitive approach. - Rostov n / a: Publishing house of the
Russian State University, 2006. – 332p.
11. Avdeeva ZK, Kovriga SV, Makarenko DI, Maksimov VI, Cognitive approach in
management, Probl. Upr., 2007, issue 3, p. 2–8.
12. Bolotova L.S. Artificial intelligence systems: models and technologies based on
knowledge: textbook / FSBEI RSUITP; FGAU GNII ITT "Informatics". - M: Finance and
statistics, 2012-- 664 p.
13. Borisov V.V., Kruglov V.V., Fedulov A.S. Fuzzy moels and networks. - 2nd ed.
stereotypes. - Hot line - Telecom, 2015. 284 p.
14. Goridko, N. P. Modern economic growth: theory and regression analysis: Monograph /
N.P. Goridko, R.M. Nizhegorodtsev. - M: Infra - M, 2017-- 444p.
15. Draper, N. Applied regression analysis / N. Draper. - M: Williams I.D., 2019.- 912p.
16. Sokolov G.A. Introduction to regression analysis and planning of regression experiments
in economics: Textbook / G.А. Sokolov, R.V. Sagiots. - M.: Infra-M, 2016, -352p.
17. Mikryukov A.A., Gasparian M.S., Karpov D.S. Development of proposals for promoting
the university in the international institutional ranking QS based on statistical analysis
methods. Statistics and Economics. 2020; 17 (1): 35-43.
18. Belyaev VI Marketing: fundamentals of theory and practice: textbook. / V.I.Belyaev. - 2nd
ed., Stereotype. - Moscow: KNORUS, 2007.- 672 p.
19. Bogomolova EV SWOT analysis: theory and practice of application / EV Bogomolova //
Economic analysis: theory and practice. - 2004. - No. 17 (32). - pp. 57-60.
20. Bogomolova EV Conceptual approach to the application of the SWOT-analysis
methodology in trade organizations to assess the effectiveness of the organization's
functioning / Е.А. V. Bogomolova // Audit and financial analysis.
21. Trakhtengerts E.A. Computer support for the formation of goals and strategies. - M.:
SINTEG, 2005-224 p.
22. Kornoushenko E.K., Maksimov V.I. Situation control using the structural properties of the
cognitive map // Tr. IPU RAS. - 2000. - T. XI. - p. 85-90.
23. Information and analytical system QS - analytics. https://analytics.qs.com/#/signin.
24. O.P. Kuznetsov, “Intellectualization of control decision support and creation of intelligent
systems”, Probl. Upr., 2009, No. 3.1, p.4–72.
25. Saati T. Decision making. Method of analysis of hierarchies: per. from English M: Radio
and communication, 1993.320 p.
26. V. V. Kulba Scenario analysis of the dynamics of behavior of socio-economic systems /
V.V. Kulba, D.A. Kononov, S.S. Kovaleskiy et al. - Moscow: IPU RAN, 2002. - 122 p.
27. Maksimov VI Structural-target analysis of the development of socio-economic situations
// Management problems. 2005. No. 3.
28. Gorelova G.V., Melnik E.V., Korovin Y.S. Cognitive analysis, synthesis, forecasting the
development of large systems of intelligent RIUS "Artificial Intelligence" 3'2010. - p. 61-
72.
29. Roberts F.S. Discrete mathematical models with applications to social, biological and
environmental problems / F.S. Roberts - M.: Nauka, 1986.
30. V. V. Kulba Methods for the formation of scenarios for the development of socio-
economic systems / V. V. Kulba, D. A. Kononov, S. A. Kosyachenko, A. N. Shubin.
Moscow: SINTEG, 2004.296 p.
31. Simon H. The Structure of Ill-structured Problems // Artificial Intelligence. - 1973. - Vol.
4. - P. 181-202.
32. Cognitive analysis and management of the development of situations: materials of the 1st
international conference. Moscow: IPU, 2001.196 p.
33. Checkland P.B. Systems Thinking, Systems Practice. - New York: Wiley, 1981.
34. Axelrod R. The Structure of Decision: Cognitive Maps of Political Elites. - Princeton:
University Press, 1976.
35. Swain J.J. Simulation Software Boldly Goes ..., // OR / MS Today. - 2007. - Vol. 36, No.
5. - P. 50-61.
36. Kosko B. Fuzzy cognitive maps // Man-Machine Studies. - 1986. Vol. 24. - P. 65-75
Eden C. Cognitive mapping // European Journal of Operational Research. - 1988. - N 36. -
P. 1-13.
37. Kizim N.A., Khaustova V.E. Features of testing models based on cognitive maps for
stability and reliability // Modern approaches to modeling complex socio - economic
systems. - Kh .: FLP Aleksandrova K. M .; Publishing House INZHEK, 2011. - 280 p.