A MULTI-CRITERIA DECISION ANALYSIS (MCDA)-BASED EVALUATION MODEL FOR PERSONALIZED LEARNING PLATFORMS IN HIGHER EDUCATION

THE VI INTERNATIONAL SCIENTIFIC CONFERENCE “SCIENTIFIC FOUNDATIONS FOR THE USE OF INFORMATION TECHNOLOGIES OF A NEW LEVEL AND MODERN PROBLEMS OF AUTOMATION”, NOVEMBER 20, 2025 201 A MULTI-CRITERIA DECISION ANALYSIS (MCDA)-BASED EVALUATION MODEL FOR PERSONALIZED LEARNING PLATFORMS IN HIGHER EDUCATION Zaripova Mukaddas Djumayozovna1, Abdilamiyeva Noila Ramiddinovna2 1Associate Professor, Department of Computer and Software Engineering Termez State University, Uzbekistan; 21st-Year Master’s Student, Major in Computer Systems and Software Engineering Termez State University, Uzbekistan https://doi.org/10.5281/zenodo.17739858 Abstract. In recent years, the rapid integration of artificial intelligence and adaptive technologies into higher education has accelerated the shift toward personalized learning. However, despite the proliferation of platforms such as ALEKS, Knewton Alta, Smart Sparrow, Realizeit, and Coursera, there remains a lack of comprehensive evaluation models that can systematically compare their pedagogical, technical, and usability effectiveness. This study proposes a Multi-Criteria Decision Analysis (MCDA)-based evaluation model to assess the overall performance of personalized learning platforms across ten key criteria, including adaptive algorithms, learning analytics, user interactivity, instructor involvement, technical integration, data security, and cost-efficiency. The research adopts a mixed-method approach, combining qualitative content analysis and quantitative scoring based on a five-point Likert scale. The collected data were analyzed through weighted aggregation to calculate the integrated efficiency index (Ip) for each platform. Findings reveal that Realizeit achieved the highest overall score (4.6/5), demonstrating strong AI-driven adaptability and superior LMS integration. ALEKS ranked second (4.1/5) due to its effective gap-analysis algorithm, while Coursera showed the lowest adaptability index (3.6/5), mainly due to limited personalization depth. The proposed MCDA-based model provides a systematic and replicable framework for decision-makers in higher education institutions to select, implement, and evaluate digital learning platforms effectively. Keywords. Personalized Learning; Adaptive Learning; Higher Education; Multi-Criteria Decision Analysis (MCDA); Learning Analytics; Artificial Intelligence; Digital Integration Introduction In recent years, the rapid advancement of artificial intelligence (AI) and digital technologies has significantly transformed the landscape of education, giving rise to a new phase in personalized learning. This approach enables the creation of individualized learning pathways tailored to each student's knowledge level, learning style, and interests. In higher education, personalized learning plays a vital role in improving academic performance, enhancing independent thinking, and fostering analytical competencies among students. The relevance of this study lies in the need to identify which personalized learning platforms currently used in higher education provide the most effective learning experiences under the conditions of digital transformation. Therefore, this research conducts a comparative analysis of widely adopted platforms (ALEKS, Knewton (Alta), Smart Sparrow, Realizeit, and Coursera) to evaluate their effectiveness across multiple dimensions, including adaptive algorithms, learning THE VI INTERNATIONAL SCIENTIFIC CONFERENCE “SCIENTIFIC FOUNDATIONS FOR THE USE OF INFORMATION TECHNOLOGIES OF A NEW LEVEL AND MODERN PROBLEMS OF AUTOMATION”, NOVEMBER 20, 2025 202 analytics capabilities, user experience (student and instructor interfaces), instructor role and pedagogical management tools, technical integration (LMS, SSO, API), and data security. Through this comparative evaluation, the study aims to identify the most suitable platforms and develop evidence-based recommendations for optimizing personalized learning processes, monitoring academic performance, supporting instructors, and implementing digital infrastructure in higher education institutions. The findings of this research are expected to support strategic digital decision-making within universities by enabling efficient resource allocation, modernization of curricula and teaching methodologies, and the design of integration roadmaps compatible with existing LMS environments. Furthermore, the proposed evaluation criteria and five-point rating framework not only highlight the relative strengths and limitations of each platform but also provide a methodologically grounded basis for higher education institutions to make contextually appropriate and data-driven choices. Literature Review In recent years, personalized learning — particularly AI-driven educational platforms — has become a central topic in higher education research. Studies conducted between 2020 and 2025 have primarily focused on three major themes: the effectiveness of adaptive learning, the pedagogical role of instructors, and the technical integration of learning systems. Collectively, these works highlight the transformative potential of adaptive learning technologies while also revealing critical gaps in their empirical validation and implementation practices. C. Merino-Campos (2025) examined the impact of AI-based personalized learning systems in higher education through a systematic review of 45 academic sources. The study found that such systems enhance student motivation and improve the precision of assessments, although challenges related to empirical validation and technical integration remain unresolved [1]. Similarly, S. Saleem et al. (2025) surveyed university instructors to explore the adoption of personalized learning tools, finding that while these systems increase student engagement, insufficient digital readiness among instructors limits their effective utilization [2]. From a digital competency perspective, H. Yaseen (2025) demonstrated that higher levels of digital literacy among students correlate with greater benefits from adaptive learning environments [3]. W. Strielkowski (2025) extended this discussion by situating adaptive learning within the framework of sustainable education, emphasizing its ability to individualize learning, enable real-time monitoring, and support data-informed teaching practices [4]. Technological advances were also explored by P. Shi et al. (2025), who applied deep learning algorithms to develop an adaptive model for teaching history. Their experimental study revealed that students using adaptive methods outperformed those in traditional instruction [5]. On the other hand, H. Harris (2024) analyzed teachers’ perceptions of AI and adaptive systems, identifying both optimism toward pedagogical efficiency and concerns about reduced instructor agency and academic integrity [6]. Across these studies, consistent findings indicate that personalized learning technologies can significantly improve student outcomes, flexibility, and analytical depth. However, methodological diversity — including surveys, meta-analyses, and experimental approaches — has led to fragmented insights across pedagogical, technical, and psychological domains. A synthesis of the reviewed literature reveals that several dimensions of personalized learning remain underexplored: Instructor Role: Current research provides limited insight into how instructors interact with THE VI INTERNATIONAL SCIENTIFIC CONFERENCE “SCIENTIFIC FOUNDATIONS FOR THE USE OF INFORMATION TECHNOLOGIES OF A NEW LEVEL AND MODERN PROBLEMS OF AUTOMATION”, NOVEMBER 20, 2025 203 adaptive learning systems and how their managerial and facilitative roles can be effectively modeled. As a result, comprehensive frameworks for assessing teacher-mediated adaptive learning remain insufficient. Technical Integration: Few studies address the interoperability of platforms with existing institutional systems such as LMS, API infrastructures, and data security protocols. This gap restricts the systematic evaluation of technological compatibility and information flow efficiency. User Experience: Research on user interface design, usability, and motivational factors for students and instructors remains scarce. Moreover, empirical assessments of how user experience directly affects learning performance are limited. Therefore, this study aims to address these gaps by conducting a comparative evaluation of personalized learning platforms in higher education - focusing on their effectiveness, technical integration, and user experience. By applying a structured, MCDA-based evaluation framework, this research seeks to contribute a holistic understanding of adaptive learning ecosystems and inform evidence-based digital transformation in higher education. Research Methodology This study employed a mixed research approach, combining elements of both qualitative and quantitative analysis. Such an approach makes it possible to comprehensively examine the effectiveness of personalized higher education platforms from multiple perspectives and according to various evaluation criteria. The data were collected from secondary sources. The main sources included scientific articles published between 2020 and 2025, analytical reviews in international journals, technical documentation of the platforms (ALEKS, Knewton Alta, Smart Sparrow, Realizeit, and Coursera), and user experience reports. In addition, open-access reports from universities that have implemented adaptive learning practices were also utilized. The research sample was formed purposefully, focusing on materials directly related to the research topic and characterized by a high degree of academic reliability — such as articles published in internationally indexed journals and official technical or user documentation of the platforms. The sample size was determined based on the analytical depth required and the relevance of the materials to the study objectives. The collected data were processed using content analysis and thematic analysis methods to identify the main research directions and patterns. It should be noted, however, that the study has several limitations. Only open-access materials published between 2020 and 2025 were analyzed; due to the lack of empirical (experiment-based) data, some conclusions are derived from secondary evidence. Furthermore, as the internal algorithms of certain platforms (for example, the code structures of ALEKS and Realizeit) are not publicly available, their technical aspects were assessed based on general analytical descriptions. Despite these limitations, the chosen approach made it possible to compare personalized higher education platforms from different perspectives and to identify their advantages and limitations in terms of pedagogical effectiveness, technical integration, and user experience. Results and Discussion The results of the study were determined through a comparative evaluation of personalized higher education platforms across ten key criteria — adaptive model/algorithm (M₁), degree of personalization (M₂), learning process analytics (M₃), learning pathway design (M₄), interactivity and user experience (M₅), instructor role and management capabilities (M₆), technical THE VI INTERNATIONAL SCIENTIFIC CONFERENCE “SCIENTIFIC FOUNDATIONS FOR THE USE OF INFORMATION TECHNOLOGIES OF A NEW LEVEL AND MODERN PROBLEMS OF AUTOMATION”, NOVEMBER 20, 2025 204 infrastructure and integration (M₇), data security (M₈), overall effectiveness (M₉), and pricing and licensing model (M₁₀). The evaluation results are presented in Table 1, which comprehensively reflects the overall performance indicators of each platform. Table 1. Evaluation results based on criteria № Criteria ALEKS Knewton Alta Smart Sparrow Realizeit Coursera M₁ Adaptive model/algorithm 5 4 4 5 3 M₂ Degree of personalization 5 4 4 5 3 M₃ Learning process analytics 4 4 4 5 3 M₄ Learning pathway 4 4 4 5 3 M₅ Interactivity and user experience 4 4 4 4 5 M₆ Instructor role and management capabilities 4 3 5 4 3 M₇ Technical infrastructure and integration 4 4 3 5 3 M₈ Data security 4 4 3 5 4 M₉ Overall effectiveness 4 4 4 5 3 M₁₀ Pricing and licensing model 3 4 3 4 5 Average score (Iₛ) ---- 4.1 3.9 3.8 4.6 3.6 When analyzing the data presented in Table 1, it was found that the Realizeit platform achieved the highest overall score of 4.6, which can be attributed to its AI-driven adaptive adjustment algorithms, real-time analytics capabilities, and a high degree of integration with LMS, API, and data security systems. The ALEKS platform ranked second with an average score of 4.1, distinguished by its high precision in diagnosing learners’ knowledge gaps and constructing individualized learning pathways. Knewton Alta (3.9) and Smart Sparrow (3.8) demonstrated moderate performance. Their strengths lie in interactive engagement with instructors and adaptive content generation, though both platforms show limitations in terms of technical integration and data protection. Coursera (3.6), despite its global effectiveness, received a comparatively lower score due to its limited level of personalization. Thus, the analysis results made it possible to systematically compare personalized learning platforms based on the Multi-Criteria Decision Analysis (MCDA) model. The scores obtained through this approach served as the basis for calculating the integrated efficiency index (Iₛ) in the subsequent stage. THE VI INTERNATIONAL SCIENTIFIC CONFERENCE “SCIENTIFIC FOUNDATIONS FOR THE USE OF INFORMATION TECHNOLOGIES OF A NEW LEVEL AND MODERN PROBLEMS OF AUTOMATION”, NOVEMBER 20, 2025 205 In this study, a multi-criteria integrated evaluation model was developed to assess personalized higher education platforms. The formation of this model is grounded in the principles of Multi-Criteria Decision Analysis (MCDA) [7,8] and the integrated indexing approach [9,10]. Through this framework, personalized learning platforms were comprehensively evaluated across pedagogical, technical, and economic indicators. The purpose of the model is to determine a single numerical value representing the overall efficiency of each platform by aggregating multiple dimensions — pedagogical, technical, interactive, and economic — into a unified performance index. The evaluation model is expressed by the following formula:  = = n i ip M n I 1 1 (1) where: Ip – the integrated efficiency index of platform p; Mi – the score for criterion i (ranging from 1 to 5); n – the total number of criteria (in this study, n=10). Each platform was evaluated using a five-point Likert scale, where 1 – poor, 2 – fair, 3 – average, 4 – good, and 5 – excellent. The mathematical essence of the model lies in the fact that each platform’s evaluations across all criteria are expressed as a normalized average value. This approach allows the comparison of different types of criteria within a unified measurement framework. Consequently, the value of Ip quantitatively reflects the overall efficiency of a platform on a scale from 1 to 5. Thus, the evaluation results presented in Table 1 serve as the input data for the proposed integrated model. The final Ip value for each platform was calculated using the MCDA approach, and these results were subsequently analyzed in depth in the sections on Scientific Novelty and Practical Significance. This model has been applied for the first time to personalized higher education platforms — ALEKS, Knewton Alta, Smart Sparrow, Realizeit, and Coursera — enabling a combined analysis from pedagogical, technical, and economic perspectives. Whereas previous studies typically focused on either pedagogical or technical aspects in isolation, the present model provides a comprehensive assessment encompassing multiple dimensions — adaptivity, learning analytics, interactivity, technical integration, data security, and economic efficiency. The practical significance of the research lies in the fact that the developed model allows higher education institutions to scientifically justify methodological and analytical decisions when selecting, implementing, and evaluating digital learning systems. The ten-criteria evaluation framework (including adaptive model, instructor role, user interface, technical integration, data security, and pricing) offers universities an objective, evidence-based set of benchmarks for platform selection. Moreover, this approach enables the formulation of recommendations aimed at improving educational quality, developing digital infrastructure, and enhancing academic performance monitoring. In this way, the study presents a comprehensive model that bridges the gap between educational theory and practical implementation in higher education. The analysis results indicate that personalized higher education platforms differ significantly from one another, and their overall efficiency largely depends on the quality of adaptive algorithms, user experience, and the degree of technical integration. The highest integrated efficiency index was observed in Realizeit (Ip = 4.6). This result can be explained by THE VI INTERNATIONAL SCIENTIFIC CONFERENCE “SCIENTIFIC FOUNDATIONS FOR THE USE OF INFORMATION TECHNOLOGIES OF A NEW LEVEL AND MODERN PROBLEMS OF AUTOMATION”, NOVEMBER 20, 2025 206 the platform’s AI-driven adaptive module, real-time learning analytics capabilities, and a robust technical infrastructure that ensures seamless integration with LMS and API systems. The ALEKS platform (Ip = 4.1) ranked second, owing to its Knowledge Space Theory–based precision in diagnosing learning gaps and its strong ability to generate individualized learning paths. Meanwhile, Knewton Alta and Smart Sparrow (Ip ≈ 3.8–3.9) demonstrated solid performance in adaptive content creation and instructor–student interaction mechanisms, although their scores were lowered by limitations in technical integration and data security. Coursera (Ip = 3.6) — as a global learning platform — reaches a wide audience; however, the relatively limited scope of its personalized learning algorithms has reduced its overall integrated efficiency. The evaluation results also revealed noticeable differences among the platforms in terms of instructor role, technical integration, and user experience. For instance, Realizeit and ALEKS provide instructors with advanced tools for active monitoring and individualized analytics, whereas Coursera and Smart Sparrow rely more heavily on automated processes, limiting realtime instructor interaction—an element that directly influences the quality of learning outcomes. Findings derived from the MCDA-based model confirm that the synergy between instructor involvement, adaptive algorithms, and technical integration serves as a key determinant in enhancing the effectiveness of personalized learning environments. These results are consistent with earlier studies ([1], [2], [8]), which demonstrated that user engagement, digital literacy, and technological adaptability have a direct impact on learning outcomes. However, the distinct contribution of the present research lies in its ability to integrate pedagogical, technical, and economic indicators into a single composite index, providing a unified evaluation framework. This integrative approach establishes a scientifically grounded decision-making tool for higher education institutions, supporting the development of digital transformation strategies, optimization of learning management systems, and data-driven enhancement of academic quality. Conclusion The findings of this study demonstrate that the pedagogical, technical, and economic aspects of personalized higher education platforms are deeply interconnected. According to the results obtained through the Multi-Criteria Decision Analysis (MCDA) – based evaluation model, the Realizeit platform achieved the highest efficiency index, while ALEKS ranked second for its accuracy in diagnosing knowledge gaps and generating individualized learning pathways. However, the analysis also revealed that Coursera, Knewton Alta, and Smart Sparrow possess certain limitations in terms of instructor involvement, data security, and user experience. The scientific conclusion derived from this study is that enhancing the effectiveness of personalized learning requires the harmonious integration of adaptive algorithms, interactive instructor engagement, technical integration, and data security mechanisms within the platforms. The MCDA-based integrated evaluation model successfully unified these diverse criteria into a single quantitative indicator, allowing for a comprehensive assessment of platform performance. The application of this model provides higher education institutions with a scientifically grounded tool for making data-driven decisions in the selection, evaluation, and implementation of digital learning systems. From a practical perspective, the model’s results can be utilized in the following ways: To standardize platform selection criteria for universities (adaptability, integration, security, cost); To evaluate the effectiveness of digital learning systems through an integrated index in quality monitoring processes; THE VI INTERNATIONAL SCIENTIFIC CONFERENCE “SCIENTIFIC FOUNDATIONS FOR THE USE OF INFORMATION TECHNOLOGIES OF A NEW LEVEL AND MODERN PROBLEMS OF AUTOMATION”, NOVEMBER 20, 2025 207 To design targeted training programs and methodological materials aimed at improving instructors’ digital competencies; To provide an analytical foundation for educational policy development (e.g., within the framework of a National Digital Education Strategy). Thus, this research is significant both theoretically — by proposing an integrated evaluation model based on MCDA — and practically, by offering a decision-making and optimization tool for the implementation and improvement of digital learning ecosystems in higher education institutions. Recommendations and Future Prospects The findings of this study indicate that enhancing the effectiveness of personalized higher education platforms requires the coordinated development of pedagogical, technical, and data security components. Accordingly, the following practical recommendations are proposed: • Pedagogical Integration: It is recommended to implement professional development programs aimed at strengthening instructors’ digital pedagogical competencies for working with personalized learning platforms. This will enhance teachers’ active participation in adaptive systems and improve their impact on students’ learning outcomes. • Technical Integration: To ensure seamless interoperability of platforms with LMS, API, and Learning Analytics systems, it is necessary to develop the national educational information infrastructure based on an integrated model. • User Experience (UX): A UX audit methodology should be designed to improve interface simplicity, visual design, and usability for both students and instructors. • Data Security: Platforms should adopt data protection policies aligned with GDPR and ISO 27001 standards, with stronger encryption and access control mechanisms for user information. • Economic Aspect: It is advisable to develop financing mechanisms based on freemium or public–private partnership models to ensure cost-effective implementation for higher education institutions. In terms of future research directions, the following perspectives are identified: • Empirical Testing: Conduct pilot studies across multiple universities using the developed MCDA model and analyze the outcomes; • AI-Driven Adaptive Analysis: Develop artificial intelligence models capable of predicting students’ performance, motivation, and learning engagement; • Cross-Platform Research: Compare digital learning platforms across different academic disciplines (engineering, social sciences, medicine, arts); • Neuro-Methodological UX Studies: Apply experimental methods such as eye-tracking, EEG, and emotion-based response analysis to evaluate user experience; • Policy-Level Application: Explore the potential integration of the proposed model into Uzbekistan’s “Digital Education 2030” strategic initiatives. In conclusion, this study introduces a novel approach that bridges theoretical foundations and practical solutions in evaluating personalized higher education systems. Future research will continue along this trajectory by integrating digital education policies with innovative technologies to strengthen adaptive and data-driven learning ecosystems. REFERENCES THE VI INTERNATIONAL SCIENTIFIC CONFERENCE “SCIENTIFIC FOUNDATIONS FOR THE USE OF INFORMATION TECHNOLOGIES OF A NEW LEVEL AND MODERN PROBLEMS OF AUTOMATION”, NOVEMBER 20, 2025 208 1. Merino-Campos, C. (2025). The Impact of Artificial Intelligence on Personalized Learning in Higher Education: A Systematic Review. Trends in Higher Education, 4(2), 17. 2. Saleem, S., Aziz, M. U., Iqbal, M. J., & Abbas, S. (2025). AI in Education: Personalized Learning Systems and Their Impact on Student Performance and Engagement. Critical Review of Social Sciences Studies, 3(1), 2445–2459. 3. Yaseen, H. (2025). The Moderating Role of Digital Literacy. Sustainability, 17(3), 1133. 4. Strielkowski, W. (2025). AI-Driven Adaptive Learning for Sustainable Educational Development. Sustainable Development Journal. 5. Shi, P., et al. (2025). Adaptive Learning Oriented Higher Educational Classroom: New Teaching Plans through Deep Learning Methods. Scientific Reports. 6. Harris, H. (2024). Faculty Perspectives Toward Artificial Intelligence in Higher Education. Doctoral dissertation, Middle Georgia University. 7. Belton, V., & Stewart, T. J. (2002). Multiple criteria decision analysis: An integrated approach. Springer Science & Business Media. https://doi.org/10.1007/978-1-4615-1495-4 8. Greco, S., Ishizaka, A., Tasiou, M., & Torrisi, G. (2019). On the methodological framework of composite indices: A review of the issues of weighting, aggregation, and robustness. Social Indicators Research, 141(1), 61–94. https://doi.org/10.1007/s11205-017-1832-9 9. Ishizaka, A., & Nemery, P. (2013). Multi-criteria decision analysis: Methods and software. John Wiley & Sons. https://doi.org/10.1002/9781118644898 10. Organisation for Economic Co-operation and Development (OECD). (2008). Handbook on constructing composite indicators: Methodology and user guide. OECD Publishing. https://doi.org/10.1787/9789264043466-en.