A MULTI-CRITERIA DECISION-MAKING APPROACH TO HUMAN RESOURCE OPTIMIZATION IN BANK

ABSTRACT

This study applies a multi-criteria decision-making (MCDM) approach to human resource management challenges in the banking sector. Among various MCDM methods, the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) is selected and customized tackle human resource management challenges in the banking sector, focusing on multi-criteria decision-making in complex and uncertain environments. The methodology builds upon established TOPSIS frameworks by incorporating expert judgments and linguistic variables to evaluate applicants based on subjective criteria. This research emphasizes the transformation of qualitative assessments into quantitative metrics, enabling a more objective and precise evaluation process. An Intelligent Decision Support System (IDSS) is developed to implement this approach, leveraging fuzzy logic to address the inherent uncertainty and imprecision of human judgment. By assigning fuzzy membership values to linguistic terms, the system captures the subtleties of expert evaluations. While inspired by existing research, this study adapts the TOPSIS method specifically for HRM tasks, demonstrating its applicability and effectiveness in ranking potential candidates in Azerbaijani banks.

АННОТАЦИЯ

В этом исследовании применяется подход многокритериального принятия решений (MCDM) к задачам управления человеческими ресурсами в банковском секторе. Среди различных методов MCDM выбран и адаптирован Метод упорядочения предпочтений по близости к идеальному решению (TOPSIS) для решения задач управления человеческими ресурсами в банковском секторе, фокусируясь на многокритериальном принятии решений в сложных и неопределенных условиях. Методология основывается на устоявшихся рамках TOPSIS, включая экспертные суждения и лингвистические переменные для оценки кандидатов на основе субъективных критериев. В этом исследовании подчеркивается преобразование качественных оценок в количественные метрики, что позволяет сделать процесс оценки более объективным и точным. Для реализации этого подхода разработана Интеллектуальная система поддержки принятия решений (IDSS), использующая нечеткую логику для устранения присущей человеческому суждению неопределенности и неточности. Присваивая нечеткие значения членства лингвистическим терминам, система улавливает тонкости экспертных оценок. Вдохновленное существующими исследованиями, данное исследование адаптирует метод TOPSIS специально для задач HRM, демонстрируя его применимость и эффективность при ранжировании потенциальных кандидатов в азербайджанских банках.

Keywords. human resource management, TOPSIS, fuzzy logic, intellectual decision support system, bank.

Ключевые слова: управление человеческими ресурсами, TOPSIS, нечеткая логика, интеллектуальная система поддержки принятия решений, банк.

1. Introduction

Strategic Human Resource Management (SHRM) serves as a cornerstone of effective human resource practices, emphasizing the alignment of HR activities such as recruitment, training, performance evaluation, and compensation with the broader strategic goals of the organization [1]. Among the various analytical tools used in human resource management, the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) is a well-established multi-criteria decision-making (MCDM) method developed by Hwang and Yoon (1981). The main idea of TOPSIS is to evaluate alternatives based on their distance from two reference points: the positive ideal solution (the best possible option) and the negative ideal solution (the worst possible option). The alternative closest to the ideal solution and farthest from the negative ideal solution is considered the most suitable. Because it balances multiple, often conflicting, criteria, TOPSIS has been widely applied in fields requiring complex evaluations, including finance, engineering, and human resource management. This alignment is particularly critical in the banking sector, where employee skills and competencies directly influence financial outcomes, operational efficiency, and customer satisfaction.

The banking sector presents unique challenges for HRM, characterized by its highly dynamic nature, strict regulatory frameworks, rising customer expectations, and rapid technological advancements [2]. Digital transformation has further heightened the importance of strategic HR practices, as banks must now identify employees with skills relevant to evolving roles, implement robust upskilling and reskilling programs, and manage workforce changes driven by automation [3, 4]. These challenges necessitate advanced decision-making frameworks that can optimize workforce performance while ensuring long-term sustainability and innovation in the sector [5].

This study adopts the TOPSIS method to address HRM challenges in Azerbaijani banks, particularly for optimizing decisions in the recruitment and evaluation of employees. TOPSIS is chosen over other MCDM methods due to its robustness in handling decision-making in fuzzy environments, where criteria values are often imprecise or linguistic. Its ability to incorporate both objective weights and subjective preferences makes it a particularly suitable choice for the multi-criteria evaluation of human resource candidates in the banking sector.

Numerous studies have investigated the application of intelligent systems in multi-criteria The application of intelligent systems for multi-criteria decision-making (MCDM) and decision support in HRM has been extensively explored, particularly in the banking domain [6–8]. These studies focus on key challenges such as managing the hierarchical structure of decision criteria, processing complex information, and incorporating expert knowledge. A significant challenge in HRM decision-making is the use of linguistic variables and imprecise data, which necessitate the adoption of fuzzy set and fuzzy logic theories for effective information management [9–13].

In Azerbaijani banks, the adoption of intelligent technologies and decision-support systems remains limited, creating significant barriers to efficient human resource management. Developing AI-driven decision-support systems that incorporate MCDM techniques is crucial for overcoming these limitations and addressing the sector's dynamic challenges. Industry 4.0 has further amplified this need, with AI-enabled Decision Support Systems (DSS) offering enhanced capabilities for managing complex decision-making scenarios within Intelligent Decision Support Systems (IDSS).

Researchers such as Linger and Burstein have proposed frameworks for IDSS that consist of a pragmatic layer focused on task performance and a conceptual layer emphasizing task structures and processes. These frameworks form the basis for building IDSS architectures, as illustrated in Fig. 1 [14, 15].

Figure 1 illustrates the architecture of an Intelligent Decision Support System (IDSS) specifically tailored for the banking sector. This architecture integrates three primary components - Model Base, Knowledge Base, and Database - to facilitate informed and efficient decision-making in human resource management (HRM).

Figure1. Architecture of IDSS in banks

The purpose of the work is to build complex decision-making systems in the bank and need effective ways to evaluate a wide range of applicants. In this article, as a method of finding the best applicant or applicants in the recruitment process in banks - TOPSIS is preferred and the solution algorithm is shown.

2. Research Methodology

Based on a complex approach to the management of human resources in the banking system in Azerbaijan, the model of decision-making in HRM tasks can be given conceptually by the following formula:

M_HRM = (X, K, Y, E, V, P, L, W)                                                       (1)

Where:

• X = {, i = }: the set of applicants from which the best one should be selected;
• K = { , j = }: the set of criteria for evaluating applicants;
•   = {  , t = }: the indicators of criteria;
• Each part of Y: the range for determining the value of the criterion;
• E: the group of experts participating in bank decision-making;
• V: the relationships between experts according to the preferences of the decision-maker in the bank;
• P: the relationship between the X, K, and E sets;
• L: linguistic expressions reflecting the level of partial criteria satisfaction by the applicant (membership level);
• W: the relationship between criteria and partial criteria.

This conceptual model aligns with approaches discussed in the peer-reviewed work of M.H. Mammadova, Z.G. Jabrayilova, 2018 [17], where a similar framework was proposed for addressing multi-criteria decision-making (MCDM) tasks in HRM. Specifically, this study adapts and extends the methodology described in [17] to accommodate the unique challenges of HRM in the Azerbaijani banking sector. By incorporating the principles of fuzzy logic and the TOPSIS method, the model effectively addresses the complexities of linguistic variables and subjective criteria evaluations.

The bank system utilizes a set of applicants X = {, i = }. The criteria for evaluating these applicants are represented by K = { , j = }. Each criterion ​ is further divided into sub-criteria, denoted as ​, where t = . The relative importance of each criterion and sub-criterion is quantified by the weights ​ and ​, respectively, where j = . The expertise level of each expert ​, where l = , is assessed and represented by the coefficient ​.

Consider that f(x) is an objective function that guarantees the selection of the best bank applicant:

1. f(x) = max(f(), f(), f(),..., f()) f(x) → [0, 1],
where f() is the result vector of the applicant's evaluation  ∈ X according to the integral K criterion.

2. K() = (p(), w, v),
where K() is the integral evaluation of applicant ​ based on:

• p(): the experts' choices using linguistic variables;
• w= (), z = 1, …., Z: the weights of the partial criteria;
• v= (),: the coefficients of relative importance of experts' skills.

3. f() > 0, provided f() > 0.

4. g(K(x), w, v) ∈ G, x ∈ X.

5.  , j = , .

6.  > 0, t = .

7.  > 0, z =  .

8. l = , .

This study applies the TOPSIS method to conceptualize an approach for addressing human resource management (HRM) challenges in the banking sector. By incorporating multi-criteria decision-making (MCDM) principles and fuzzy logic, the proposed model offers a structured framework for evaluating and ranking applicants based on subjective criteria and expert opinions. The model is grounded in theoretical exploration and demonstrates the feasibility of integrating the TOPSIS method within Intelligent Decision Support Systems (IDSS) for HRM.

However, due to the exploratory nature of this research, experimental validation and implementation of the proposed methodology in a real-world banking environment were not conducted. Future work will focus on practical experimentation and testing to evaluate the effectiveness and reliability of the model in decision-making scenarios.

3. Discussion of the results obtained

The TOPSIS method relies on numerical data to rank alternatives. To accommodate the qualitative nature of expert judgments, linguistic variables are transformed into fuzzy trapezoidal numbers. This representation captures the uncertainty inherent in human judgment and provides a more accurate representation of the decision-maker's preferences.

The operations of addition, subtraction, and multiplication of fuzzy trapezoidal numbers  = (, , , ) and  = (, , , )are defined as follows. These operations are essential for performing calculations with fuzzy numbers, which are used to represent uncertain or imprecise information.

 ⊕  = [],

   = [],

 ⊗   ≅ [],

 ⊗   = [],                                                                      (2)

 ÷   ≅

max(, ) = (max(), max(), max(), max())

min(, ) = (min(), min(), min(), min())

The distance between two fuzzy trapezoidal numbers is calculated using the following formula [13, 15]:

 (3)

To facilitate the application of numerical techniques, linguistic variables are used to represent the qualitative assessments of attributes. A seven-point linguistic scale is established, where each linguistic term corresponds to a specific range of numerical values. The Fig. 2 provides a visual representation of this mapping, illustrating how linguistic judgments are transformed into quantitative data and Table. 1 Linguistic values and fuzzy numbers.

Figure 2. Transformation of linguistic values into fuzzy trapezoidal numbers [17].

Table 1.

Linguistic values and fuzzy numbers

The purpose of setting up HR in the bank with this system is to rank the applicants based on the experts' assessments, taking into account the decision-making levels of the experts. There are several steps to solving the task:

1. In order to perform TOPSIS-based multi-objective optimization of HRM tasks in a bank, it is first necessary to extract criteria from the hierarchical structure. For this purpose, the relative importance coefficients of the criteria {,  j = }  and the partial criteria {, t = }are determined based on Saaty's AHP [18 , 19]. In the formalization, the weight of the partial criterion –  jt is the integral criterion K = { ,  j = }, that is,  = ·, where  in the calculation is  and where  is determined by multiplying . As a result, two-level hierarchical structure of selection criteria K = {,  j = } characterizing bank applicants {, t = }, which allows to cancel the hierarchical structure. Next, all the partial criteria are combined into a single G in order to simplify the indices.

G = {,  j = , t = } = {, z = }, z = +t, j = , t = ,              (4)

The parameter Z represents the total number of sub-criteria that are used to assess the suitability of applicants for various positions within the bank

2. The degree to which each applicant meets the requirements of a specific criterion is assessed using a linguistic scale (Table 1) and represented by trapezoidal fuzzy numbers. These numbers, denoted as  = ( ) = (, , , ), quantify the level of satisfaction of the criterion by the applicant. For instance, a bank expert's rating of 'good' for applicant  on criterion  is expressed as  = (7, 8, 8, 9), while a 'very good' rating is represented as  = (8, 9, 10, 10). The expert evaluations are aggregated in the following matrix:

 ↔ {, , , }, l =                                                                 (5)

3. Prior to the evaluation process, the competence level of each expert ,  l =  is assessed. These competence coefficients are then integrated into the evaluation framework to form the matrix = [] , l = , ↔ {, , , }, l = . The elements of this matrix, represented as trapezoidal fuzzy numbers, quantify the degree to which each applicant meets the requirements of a specific criterion, taking into account the expert's competence level. The elements are calculated as follows:

 = ;  = ;  = ;  =                           (6)

4. This step involves the aggregation of the individual matrices into a consolidated matrix

= [] ↔ {, , , }, l =  → ↔ {} (7)

The values of the matrix elements are computed as follows:

;

 = ;                                                                                         (8)

 = ;

;

5. Each element of matrix ↔ {} is multiplied by its corresponding criterion weight to obtain the weighted fuzzy matrix = [] ↔ {, , , }. This weighted matrix incorporates the relative importance of each criterion in the overall decision-making process:

  =

 =                                                                                           (9)

 =

 =

6. The normalization of the matrix is performed using the Hsu and Chen method [20]:

 = [] ↔ [] = [] = {, , , }                                     (10)

7. The positive ideal solution (PIS) X* is derived by considering the weighted values of each criterion for  ,  z = .

 =  = {} = {}       (11)

is selected, and the:

= [] = () = ()                                (12)

8. The negative ideal solution (NIS) is obtained by identifying the worst possible value for each criterion , where z =

 =  = {} = {}     (13)

is selected, and the:

= [] = () = ()                              (14)

9. The distance metric defined by formula (2) is employed to calculate the distance between each applicant and the PIS for each criterion.

() =  (15)

The vector  = [, ..., ] is formed by combining the individual distances of each applicant from the positive ideal solution (PIS).

10. The distance between each applicant and the NIS is calculated for each criterion

() =  (16)

The vector  = [, ..., ] is formed by combining the individual distances of each applicant from the negative ideal solution (NIS).

11. The distance between each applicant and the PIS is computed

() =                                                                     (17)

12. The distance between each applicant and the NIS is computed

() =                                                                   (18)

13. The proximity coefficient for each applicant is calculated by dividing the distance between the applicant and the negative ideal solution by the sum of the distances between the best and worst solutions. This coefficient provides a measure of how close each applicant is to the ideal solution

D() = () + ()                                                                         (19)

φ() =                                                                                        (20)

A higher proximity coefficient φ() implies a greater proximity to the ideal solution, making the applicant more desirable.

The proposed methodology leverages the TOPSIS method to address the inherent complexities of HRM tasks, particularly in the banking sector. While the conceptual framework highlights the potential advantages of this approach—such as handling subjective criteria and uncertainty through fuzzy logic—the absence of experimental validation represents a significant limitation of this study.

The lack of practical implementation means that the actual effectiveness of the method in ranking candidates and improving decision-making processes remains undetermined. To address this, subsequent research should focus on designing and conducting experiments within real-world HRM systems in banks. This would enable a comprehensive assessment of the methodology’s performance, including accuracy, efficiency, and stakeholder satisfaction.

5. Conclusion

This research applies and adapts the TOPSIS method within a fuzzy logic framework to address human resource management challenges in the banking sector, with a specific focus on Azerbaijani banks. The proposed methodology incorporates expert judgments and linguistic variables, offering a structured approach to evaluate and rank applicants based on multiple subjective and objective criteria. By transforming qualitative assessments into quantitative values, the model provides a transparent and effective means of decision-making, particularly in environments characterized by uncertainty and complexity. The integration of this approach within an Intelligent Decision Support System (IDSS) highlights its potential to enhance the efficiency and accuracy of HRM processes in banks. However, as this study is primarily theoretical, experiments to validate the methodology in real-world settings have not yet been conducted. Future research should focus on practical implementation to evaluate the effectiveness of the proposed model in banking environments. Further work may also explore integrating advanced techniques, such as machine learning and artificial intelligence, to improve predictive accuracy and adaptability. Additionally, sensitivity analyses and alternative weighting schemes should be investigated to enhance the robustness and reliability of the decision-making process.

References:

1. Gothami S. Jayarathna, Shanika R. Perera, Sanduni Gunawardana, Strategic Human Resource Management: Driving Organizational Excellence Through Strategic Alignment, Ethical Practices, and Emerging Trends, Reference Module in Social Sciences, Elsevier, 2024.
2. Saloni Pahuja, Seema Mahlawat, Vimal Kumar, Raj Kumar Sah, Minakshi Paliwal, Sumanjeet Singh, Manoj Kumar. Gaining competitive advantage status through human resource practices: A study of Indian banks. Social Sciences & Humanities Open, Volume 9, 2024,100804, ISSN 2590-2911. https://doi.org/10.1016/j.ssaho.2024.100804.
3. Nur Syhuhada Ab Ghani, Maslin Masrom,(2014). Human Resource Management Practice Models Applied in Banking Industry: A Review.
4. Ishak, N. B., Eze, U. C., & Ling, L. S. (2010). Integrating Knowledge Management and Human Resource Management for Sustainable Performance. Journal of Organizational Knowledge Management, 2010, 1–13.
5. Lina Dong, Mu Zhang. Commercial Bank Data Asset Quality Evaluation Model Based on Fermatean Fuzzy TOPSIS. Procedia Computer Science, Volume 221, 2023, Pages 565-572, ISSN 1877-0509. https://doi.org/10.1016/j.procs.2023.08.023.
6. Greshilov, A., (2014), Mathematical Methods of Decision-making, Bauman Moscow State Technical University, Moscow.
7. Mammadova, M.G, (1997), Decision-making, Based on a Knowledge Database with a Fuzzy Relational Structure, Elm press, Baku.
8. Mikoni, S., (2015), The Theory of Decision-making Management, “Lan” Press.
9. Zongmin, L., Xuedong, L., Haidong, Y., (2017), Multi-criteria group decision making method for elevator safety evaluation with hesitant fuzzy judgments, Appl. Comput. Math., 16(3), pp.296-312.
10. Kofman, A., (1982), Introduction to the fuzzy set theory, Radio and communication press, Moscow.
11. Zadeh, L.A., (1965), Fuzzy sets, Information and control, 8(3), pp.338-353.
12. Zadeh, L.A., (1996), Fuzzy logic computing with words, IEEE Transactions on Fuzzy Systems, 4(2), pp.103-111.
13. Chen, C.T., (2000), Extensions of the TOPSIS for group decision-making under fuzzy environment, Fuzzy Sets Systems, 114(1), pp.1-9.
14. Linger, H., Burstein, F., 1997. Intelligent decision support in the context of the modern organisation. In: Proceedings of the 4th conference of the international society for decision support systems – ISDSS’97, Lausanne, Switzerland, July 21–22.
15. Jahantigh, M.A., Hajighasemi, S., (2012), Ranking of generalized fuzzy numbers using distance measure and similarity measure, Int. J. Industrial Mathematics, 4(4), pp.405-416.
16. Verma Prikshat, Mohammad Islam, Parth Patel, Ashish Malik, Pawan Budhwar, Suraksha Gupta. AI-Augmented HRM: Literature review and a proposed multilevel framework for future research. Technological Forecasting and Social Change, Volume 193, 2023, 122645, ISSN 0040-1625, https://doi.org/10.1016/j.techfore.2023.122645.
17. M.H. Mammadova, Z.G. Jabrayilova. Decision-Making Support In Human Resource Management Based On Multi-Objective Optimization. TWMS J. Pure Appl. Math., V.9, N.1, 2018, pp.52-72.
18. [18] Rezaei, J., (2015), Best-worst multi-criteria decision-making method, Omega, 53, pp.49-57.
19. Saaty, T.L., (1993), Decision Making with the Analytic Hierarchy Process, Radio and communication press, Moscow.
20. Hsu, H.M., Chen, C.T., (1997), Fuzzy credibility relation method for multiple criteria decision-making problems, Information Sciences, 96(1-2), pp.79-91.