Processing a large amount of educational material requires significant time resources, which prompted us to analyze educational and cognitive activities in the framework of lifelong learning сoncept and consider practical tasks regarding forming an adaptive path of a student's learning session, with consideration for each student's unique characteristics, learning goals, and available time reserves. We consider the task of forming an adaptive learning path as a task of optimizing the student's educational-cognitive algorithm. For the first time, it was proposed to maximize the Ffed index of probability of answer without errors to one question of final control test with restrictions on execution time. The developed information technology allows you to obtain input data for evaluating student activity algorithms, evaluate learning quality indicators, and create an adaptive path for the learning session depending on specific student characteristics, educational goals, and learning time limitations. E-learning, adaptation, optimization, learning path, individual characteristics of the student, open learning resources, in particular Massive Open Online Course (Coursera, EdX, Udacity, Udemy). CMiGIN 2022: 2nd International Conference on Conflict Management in Global Information Networks, November 30, 2022, Kyiv, Ukraine ORCID: 0000-0002-5439-8750 (N. Barchenko); 0000-0002-9785-9975 (A. Tolbatov); 0000-0002-7144-7059 (T. Lavryk); 0000-0002-65649658 (V. Tolbatov); 0000-0002-8539-1252 (V. Obodyak); 0000-0001-5261-4432 (V. Yakovliev); 0000-0003-0368-8042 (Y. Motorin); 00000002-9875-7372 (Y. Artamonov)
cybersecurity, information security technologies
Professionals of cybersecurity work for various companies to protect organizations from cyberattacks and data breaches. Today, statistics show that the demand for professionals of cybersecurity is growing at a rapid rate. Therefore, the training of cybersecurity specialists today is an urgent issue both for higher educational institutions and for companies. But to prepare a universal professional of cybersecurity is a difficult task. So, most of the students, in addition to studying at the university, use external resources for the self-study of cybersecurity. This is facilitated by the large number of available Learning independently, you can get certificates for the study of a particular discipline, as well as diplomas in the specialties. A high-level specialist must constantly update his knowledge and skills throughout his life, receive new certificates of conformity and confirm existing ones. For example, cybersecurity certificates CISSP, CISA, and CISM require regular confirmation.
Processing a large amount of learning material requires significant time resources. For a working specialist, the problem of effective time allocation arises. Building information and educational systems vobodyak@id.sumdu.edu.ua (V.
Obodyak);
2022 Copyright for this paper by its authors. due to adaptive learning management algorithms will enable pupils to accomplish the desired outcomes using the resources at their disposal.
The creation of the learning session's adaptive path considering the goals of preparation, specific characteristics of the person studying (hereinafter the student), and the available time will allow plan ing learning activities rationally. Formation of creation of adaptive learning path with the help of the elearning system (ELS) was considered in the works of many scientists. However, in general, it cannot be considered resolved to the end.
Research aims to optimize models of the adaptive path formation during learning session in the ELS.
In ELS, the learning material is presented in the form of a sequence of separate electronic learning modules (ELM). Each module can be divided into parts (submodules), depending on the selected levels of complexity of the learning material. By the learning path we will understand the sequence of ELMs and self-control procedures. Various training organization options are achieved with the help of different types of self-control. By self-control, we mean the test procedure that the student completes after doing a part of the course. The number of times of self-checks is not limited. The results of selfcontrol are not taken into account by the system in the final assessment and are only a reflection of the current learning outcomes. Various types of self-control (self-control and re-studying the material, selfcontrol and extra research focusing primarily on challenging aspects of the course materials, etc.) provide a different level of quality of preparation for the final test and need the setting aside of certain time reserves.
The adaptive learning path is formed considering each student's unique qualities and available time reserve for doing ELM. Individual characteristics are taken into account when preparing the initial data.
The issue can be described generally in the following manner: it should be formed specific variant of the learning path that will provide the maximum level of assessment of the quality of learning ( ), taking into account the limitations on the learning time reserve 0, taking into account the level of complexity of the learning material 0.
We reduce this problem to the linear programming problem. When the j-th option of self-control is chosen for the i-th ELM the variable’s value will be 1. The variable's value will be 0 in all other cases. Here, represents the number of ELMs, and stands for the possible options for organizing self-control, resulting in = (̅1̅̅,̅̅) and = (̅1̅̅,̅̅̅). Let’s present the task in the following way:
P( X ) max, (1)
Thus, we have a problem in which for each ELM in the learning path only one variant of the organization of self-control is determined.
When forming the objective function for ELS, the level of assessment of the quality of learning ( ), which is measured in points, must be raised as much as possible.
Important tasks are: 1. provide for the estimation of probability-time characteristics of the learning process by input data; 2. examine the potential for constructing adaptive learning pathways in the ELS using the established model.
The issues of organizing e-learning should be considered as a complex task for the solution of which various approaches are applied:
T ( X ) T0,
U ( X ) U 0 ,
m
xij 1
j1
xij {0,1}
(2)
(3)
(4)
ergonomic (e.g, from the point of view of interaction between man and machine considering
person’s individual psychophysiological parameters) [
The key issue for e-learning systems is to offer courses that are customized for various students with varying levels of knowledge and learning rates. These systems need to be adaptable and efficient.
Basic objective of e-learning technologies realizes in provision of students with training courses that according to his individual parameters (learning rate, knowledge level). Therefore, e-learning systems must also be adaptive. A significant part of modern research emphasizes the need for adaptive e-learning and systems that can provide the conditions for such learning. For example, Carchiolo V., Longheu A. and Malgeri M. propose one of the approaches to constructng an adaptive e-learning system [7]. The authors consider an e-learning system in which adaptability is based on the student profile and teacher profile.
The authors of [8], [19], [20] propose a scheme of selection a learning object by each student individually.
Zhao and Wan proposed to implement an adaptive system [21], [22] for e-learning based on the algorithm for choosing the shortest learning paths. The authors determined that the most time- and effort-efficient learning procedure was the optimum learning path [9].
A study on the variables influencing how differently students learn when employing adaptive elearning tools was conducted, according to the authors of [10].
The optimization model will be based on the complexity of the ELM and various models of selfmonitoring processes on the construction of the adaptive path of student learning in the e-learning system, which is based on the idea of the shortest learning path algorithm [9].
Moreover, the following mathematical tools were used for solving the issue of creating a learning
path: graph theory, Petri nets [11], decision trees [12], machine learning [13, 14], fuzzy logic and others.
In [15–22], the development of the learning path was viewed as an optimization problem. Despite that,
specified studies did not consider a student's personal psychophysiological indications like cognitive
comfort [
In [
In [
Procedure - test self-control (SC)
SC 1 corresponds to the consistent development of educational material without self-control of learning outcomes.
SC 2 corresponds to the consistent development of educational material and the control procedure. If the test result is at a low level, the topic is re-studied.
SC 3 corresponds to the consistent development of educational material and the control procedure. If the test result is at a low level, only the problematic fragments of the topic are studied.
There are different ways of organizing learning that can be implemented through different models of self-control procedures. Possible options for self-control are given in Table 2.
Some possible options for the learning path with the number of difficulty levels k = 3, corresponding to three submodules, are shown in Figure 1. The learning path options differ in the quality of the learning outcomes and the time resources that are spent. To assess these indicators, the qualimetric method is used [18].
The level of complexity determines the structure of the ELM. For example, with a complexity level of = 3, there are three distinct submodules that correspond to the exploration of the subject matter at a basic, intermediate, and high levels. The choice of difficulty level is determined by the goals of the student. If it is necessary to quickly obtain a basic level of knowledge, then the path is formed with a submodule of the basic level ( = 1).
To obtain intermediate level knowledge, a path is formed with submodules of the first and second difficulty levels ( = 2).
If it is necessary to obtain high-level knowledge, then the path is formed with a submodule of all levels of complexity.
An example of the choice of components of an adaptive learning path for two ELM with different levels of complexity is shown in Figure 2.
Thus, the sequence of actions is as follows: select submodules for the learning path that correspond to the desired level of complexity of the module; check the quality of learning material, apply selfcontrol procedure options that will provide the necessary level of quality and satisfy the time limit for learning.
Therefore, for the purpose of solving the problem it is important to choose the level of complexity, generate input data for assessing the quality indicators of the path options and conduct optimization. 4.2.
The solution to the task of developing an adaptive ELS’s learning path provides an assessment of
the quality and reliability of the options for organizing the learning process. To assess the characteristics
of the quality and reliability of the learning process in ELS related to probability and time, the models
and methods described in [
The parameters that mainly determine the probabilistic-temporal indicators of the quality of ELS’s activity of the student are the level of motivation, the results of input control, functional state, cognitive comfort and time reserve. Statistics compiled from observations of student activities in the ELS allows you to obtain data for analysis and forecasting - how similar students acted in similar conditions.
The developed intelligent models [
To be able to apply optimization methods, we will develop a new objective function that allows you to maximize the value of the learning quality assessment in points ( ).
Let pij be the probability of the answer without error to one question of the final test control of the i-th ELM. In this case, if the number of ELMs is n and the number of options for organizing self-control is m the value of i and j will be accordingly = (̅1̅̅,̅̅) and = (̅1̅̅,̅̅̅). As well, is structure of the control procedure for summing up the learning outcomes (number of questions) for the i-th ELM; is the number of points that are calculated for one correct answer to the question, which corresponds to the i-th ELM; ∙ ∙ is the mathematical expectation of the total quantity of points awarded for correct responses for the final self-control that corresponds to the i-th ELM; is the mathematical expectation of the studying time for the i-th ELM with the j-th self-control option, and 0 is the highest allowable level of complexity.
Then, the problem of forming an adaptive path at a certain i-th learning step can be formulated as follows: n m ( pij xij ) gi si max i1 j1 n m Tij xij T0 i1 j1 n xij U 0 i1 n xij 1 j1 xij {0,1}
Let 1 = 20, 2 = 20, 3 = 20, 1 = 1, 2 = 1, 3 = 1, that is, the test self-control contains 20 questions for each submodule, and for each correct answer one point is awarded.
The outcomes of addressing the input data optimization problem (Table 3) with different reserves of time 0 are shown in Table 4.
The learning paths shown in Figure 3 correspond to the obtained matrices.
The 1 path is proposed for small reserves of time, offers less knowledge and no means of exercising self-control.
The solving of the issue (6) - (9) enables to identify the ideal plan X for organizing self-control of the ELM and To foresee the outcome of achieving total final control.
Consider the problem of forming an adaptive learning path of the ELM’s third complexity level ( = 3). Table 2 presents possible options for self-control of submodules.
Calculation results of the probability-time quality indicators for the options for organizing selfcontrol from Table 1 according to the method [18] are presented in Table 3. These results are input data for the optimization problem.
Although the 2 path takes more time, it offers the chance to modify the problem-learned blocks of educational content and raises the standard of learning.
Due to the model of recurrent studying in cases of poor self-control, the 13 path offers the maximum level of preparation quality.
The corresponding learning paths in Figure 4.
The developed information technology was implemented as part of the program and was researched to provide a solution to the issue of creating an adaptive learning path when studying the discipline "Information Security Technologies" for the preparation of bachelors in the specialty "Cybersecurity".
When developing the software package for the agent-manager of e-learning assistance, the specified methodology was used. Software package’s main tasks are: determination of the values of individual psychophysiological parameters of students that affect the level of cognitive and educational work; development of an adaptive learning path of the learning session, weighing the ELS's capabilities as well as the unique traits of each student.
To study the effectiveness, experiments were conducted based on Sumy State University.
Figure 5 shows an example of the use of an agent-manager for forming an adaptive path when studying the electronic learning course "Information Security Technologies" for applicants of higher education with a specialty 125 "Cybersecurity" of a bachelor's educational level.
The system provides an option for the formation of an adaptive path without self-control. Such a model slightly reduces the learning time, but significantly affects the quality of assimilation of the material. In some cases of low student motivation and significant time constraints, such a model may be appropriate.
The developed model allows us to implement the task of forming an adaptive path of a learning session in the ELS. It provides maximization of the evaluation of the learning process's quality in points, with consideration of the students' individual psychophysiological parameters and restrictions on the available learning time.
The development advantages are that, in contrast to the well-known methods for optimizing human activity algorithms, which are designed for averaged indicators, models provide an account of individual characteristics at the stage of input data formation.
It was used the indicator of probability of error-free execution as opposed to the common optimization models. A proposed indication offers the chance to get a point-based evaluation of the learning's level of quality.
Possible disadvantages and limitations of the method include the following: the possibility of use only for algorithmized discrete activities in modular ELS; the assumption of the invariability of the indicators of learning quality during work with ELM that are probabilistic-temporal.
The urgent task of creating adaptive learning path for an ELS learning session is completed.
Proposed approach allows considering the student's unique characteristics during formation of an adaptive learning path when there is a limitation on the time of learning in the ELS and taking into account the goals of the student and the complexity of the module which makes scientific novelty of the results.
Development of information technology to support e-learning makes practical value of the results. Experimental results allow us to offer developed software for learning specialists.
Future studies could address the issue of creating an adaptive path that consider the actual learning outcomes in the ELS.
As you know, actual results may differ from forecast results. This discrepancy should be considered in the model for optimizing the formation of the path. The path correction should occur each time the actual results of quality and learning time are received.
[7] V. Carchiolo, A. Longheu, M. Malgeri, Adaptive Formative Paths in a Web-based Learning Environment, Journal of Educational Technology & Society 5(4) (2002) 64–75.URL: https://www.jstor.org/stable/jeductechsoci.5.4.64. [8] M. Alian, R. Jabri, A Shortest Adaptive Learning Path in eLearning Systems: Mathematical View, Journal of American Science 5(6), 2009, 32-42. URL: https://www.jofamericanscience.org/journals/amsci/0506/07_1013_mathematical_view_am0506.pdf. [9] C. Zhao, L. Wan, A Shortest Learning Path Selection Algorithm in E-learning, in: Proceedings of the Sixth IEEE International Conference on Advanced Learning Technologies (ICALT'06), Kerkrade, Netherlands, 2006, pp. 94-95, doi: 10.1109/ICALT.2006.1652375. [10] J. R. van Seters, M. A. Ossevoort, J. Tramper, M. J. Goedhart, The influence of student characteristics on the use of adaptive e-learning material, Computers & Education 58(3) (2012) 942-952. doi: https://doi.org/10.1016/j.compedu.2011.11.002. [11] J. Dai, G. Su, Y. Sun, S. Ye, P. Liao, Y. Sun, Application of advanced Petri net in personalized learning, in: Proceedings of the 9th International Conference on E-Education, E-Business, EManagement and E-Learning, New York, NY, USA, 2018, pp. 1–6. doi: 10.1145/3183586.3183588. [12] Y.-H. Chen, C.-H. Tseng, C.-L. Huang, L. Y. Deng, W.-C. Lee, Recommendation system based on rule-space model of two-phase blue-red tree and optimized learning path with multimedia learning and cognitive assessment evaluation, Multimedia Tools and Applications 76(18) (2016) 18237–18264. doi: 10.1007/s11042-016-3717-3. [13] A. Dovbysh, I. Shelehov, S. Pylypenko, O. Berest, Estimation of Informativeness of Recognition Signs at Extreme Information Machine Learning of Knowledge Control System, CEUR Workshop Proceedings 2362 (2019) 143-152. URL: https://ceur-ws.org/Vol-2362/paper13.pdf. [14] A. Dovbysh, D. Velykodnyi, I. Shelehov, M. Bibyk, Information-extreme machine learning of the control system over the power unit of a thermal power main line, Eastern-European Journal of Enterprise Technologies 5(89) (2017) 17-24. doi: https://doi.org/10.15587/17294061.2017.112121. [15] F. Wang, L. Zhang, X. Chen, Z. Wang, X. Xu, Research on Personalized Learning Path Discovery Based on Differential Evolution Algorithm and Knowledge Graph, in: Data Science, Singapore, 2020, pp. 285–295. doi: 10.1007/978-981-15-2810-1_28. [16] A. H. Nabizadeh, J. P. Leal, H. N. Rafsanjani, R. R. Shah, Learning path personalization and recommendation methods: A survey of the state-of-the-art, Expert Systems with Applications 159 (2020). doi: 10.1016/j.eswa.2020.113596. [17] M. Ginda, M. C. Richey, M. Cousino, K. Börner, Visualizing learner engagement, performance, and trajectories to evaluate and optimize online course design, PLoS ONE 14(5) (2019). doi: 10.1371/journal.pone.0215964. [18] A. Adamenko, A. Asherov, I. Berdnikov, Informacionno-upravlyayushhie cheloveko-mashinnye sistemy: Issledovanie, proektirovanie, ispytaniya: Spravochnik, Mashinostroenie, 1993. [19] O. Viunenko, A. Tolbatov, S. Vyganyaylo, V. Tolbatov, S. Agadzhanova, S. Tolbatov, Information technologies in the educational process as the basis of modern distance learning, in: Proceedings of the 13th International Conference on Modern Problems of Radio Engineering, Telecommunications and Computer Science (TCSET), Lviv, Ukraine, 2016, pp. 831-833. doi: 10.1109/TCSET.2016.7452198. [20] S. Agadzhanova, A. Tolbatov, O. Viunenko, O. Tolbatova, Using cloud technologies based on intelligent agent-managers to build personal academic environments in e-learning system, in: Proceedings of the 2nd International Conference on Advanced Information and Communication Technologies (AICT), Lviv, Ukraine, 2017, pp. 92-96. doi: 10.1109/AIACT.2017.8020073. [21] O. Zaritskiy et al., Theoretical bases, methods and technologies of development of the professional activity analytical estimation intellectual systems, in: Proceedings of the 2nd International Conference on Advanced Information and Communication Technologies (AICT), Lviv, Ukraine, 2017, pp. 101-104. doi: 10.1109/AIACT.2017.8020075. [22] O. Zaritskiy, P. Pavlenko, A. Tolbatov in: Proceedings of the 13th International Conference on Modern Problems of Radio Engineering, Telecommunications and Computer Science (TCSET), Lviv, Ukraine, 2016, pp. 718-720. doi: 10.1109/TCSET.2016.7452162.