Multi-criteria decision-making (MCDM) is a technique that had originally been developed from the specialty of operations research, and is now a broad methodological framework applicable to complex decision support in practically all of the scientific and engineering disciplines. In most cases, when many criteria, which usually conflict with each other, should be considered at the same time, the challenge is to systematically evaluate and rank the alternatives, which is addressed by the MCDM frameworks (Taherdoost and Madanchian, 2023). Unlike single criterion optimization, MCDM recognizes that, in practice, decision making is a matter of compromises between economic, environmental, social, technical, and political aspects which cannot be reduced to one objective function. Since the early studies of Saaty (1980) on Analytic Hierarchy Process (AHP) and Hwang and Yoon (1981) on Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), there has been a tremendous development in the field of methods. The most recent MCDM models include dozens of different methods for weighting the criteria, ranking the alternatives, and dealing with uncertainties, which have their own theoretical assumptions, computational needs, and domain appropriateness Ayan et al. (2023). This diversity, however, also enriches the methodological toolbox and has made for a paradox of choice: practitioners find it more difficult to choose the right frameworks in each of their specific decision contexts.

Though extensive research has given rise to many reviews and surveys, the methodological innovation has grown at an unprecedented rate in the period of 2021–2026, especially in the field of hybrid models, high-level fuzzy extensions, and incorporating artificial intelligence. There have also been the introduction of novel methods such as BHARAT method (Rao, 2023), CRITID Zhang, et al. (2024), SITW (Kizielewicz and Sałabun, 2024), INCOME Kizielewicz et al., (2024) and LLM-based frameworks Wang, et al., (2025) as well as a lot of hybridization of existing methods and an extension in new application fields. The fast development of these requires an updated comprehensive synthesis of MCDM frameworks.

This SLR aims to answer the following research questions:

RQ1: How will methodological approaches, type of studies, and application domain of MCDM framework studies be distributed in the period 2021-26?

RQ2: What’s the most common combination of MCDM methods and extensions to fuzzy sets in hybrid approaches?

RQ3: What are the new progressions in the development of the MCDM framework, especially in the field of integrating AI, conducting dynamic decision-making and providing support for group decision-making?

RQ4: What gaps in the existing literature and/or research needs would be identified?

This review was done using the guidelines for MCDM framework review as per PRISMA 2020 Page et al. (2021) to achieve transparency, reproducibility, and all the relevant literature for MCDM framework is covered.

Several recent systematic reviews have tried to chart the MCDM landscape a different way from different angles. Basílio et al. (2022) did a thorough bibliometric study of 23,494 papers from 131 countries published between 1977 and 2022 to find out that AHP is the most widely used method worldwide. According to their analysis, the scientific production in this field grows at 14.18% per year, and China (18.50%), India (10.62%) and Iran (7.75%) are the top three countries in terms of the number of publications. The study also found three journals – Expert Systems with Applications, Sustainability and the Journal of Cleaner Production – to be the highest-ranking journals for publishing.

Ayan et al. (2023) gave an overview of new weighting techniques such as CILOS, IDOCRIW, FUCOM, LBWA, SAPEVO-M and MEREC. They analyzed them using bibliometric and content analyses to find out the trends, research components, application areas, fuzzy implementations, and hybrid studies related to each method. The results of their study revealed the increasing relevance of objective and semi-objective weighting techniques in combating the subjectivity of the conventional pair-wise comparison technique.

From the application point of view, Gebre, Cattrysse, (Alemayehu and Van Orshoven, 2021) examined the applications of MCDM methods in land allocation problems in rural areas, and concluded that methods are being used more often in Europe and China, especially AHP in ecotourism and ecosystem management. Zolghadr-Asli et al. (2021) reviewed more than 320 articles from 57 countries covering 20 years of multi-attribute decision making applications in environmental and water resources planning. Their review highlighted the importance of dealing with uncertainty using sensitivity analysis and frameworks of probability/fuzziness.

Other methodologically-oriented reviews have also been instrumental to the field. Boix-Cots, et al. (2023) have proposed a systematic classification scheme of multi-criteria group decision making methods by means of the weight aggregation techniques, which offers a general framework and expert guidance. In their systematic review of sensitivity analysis approaches in MCDA, (Wi3cki and Salabun, 2023) highlight that performing sensitivity analysis leads to more credible decision results by offering insight into possible changes that can occur in decision variants. Yuan et al. (2022) provided a review of the use of MCDA in the field of rural spatial sustainability assessment, suggesting a need for further systematic review from multi-disciplinary perspectives and for implementation studies.

Balasbaneh et al. (2025) carried out a systematic review of the literature on the MCDM approaches to the circular economy that involved checking 31 peer-reviewed publications. They discovered that MCDM is a good method for balancing environmental, economic and social aspects in assessments of circular economy, however they could not find any patterns for the selection of specific methods. A comprehensive overview of the MCDM concept, its applications, major types and techniques has been given by (Taherdoost and Madanchian, 2023). In addition, (Madanchian and Taherdoost, 2023) gave a detailed guide specifically on TOPSIS method. (Li and Hu 2021) have reviewed the multi-attribute decision-making models developed for offshore oil and gas facilities decommissioning, identifying there a need for an improved model, which considers data gaps, as well as incomplete MCDA models. A systematic literature review (SLR) was carried out by (Alexe and Nechita, 2025), who identified 49 human resource factors for R&D project success.

Although these are very valuable contributions, however, there is not yet any review that has comprehensively summarised all the developments in the MCDM framework from 2021-2026, including methodological innovations and the trends of application. This review is filling this gap by presenting a PRISMA compliant analysis of 92 studies for all aspects of MCDM.

This systematic review of 92 MCDM framework studies (2021–2026) shows a dynamic and very active field with following features:

i. First, methodological clustering around hybrid approaches in which subjective/objective weighting and distance-based, outranking, compromise and utility-based weighting are used.

ii. Sophisticated uncertainty modelling with higher order fuzzy sets (intuitionistic, Pythagorean, Fermatean, picture, spherical, neutrosophic) and other uncertainty modelling frameworks (gray system, rough set, evidential reasoning).

iii. Integration of 3 key forms of AI—machine learning for weight learning, deep reinforcement learning for sequential decisions, and large language models for end-to-end MCDM automation.

iv. Growth in application areas, stimulated by sustainability needs, led by energy, circular economy and green supply chain.

v. Lack of explicit guidance for method selection, dynamic decision-making and group decisions, validation standards and explainability.

Future Research Directions of Priority

i. A principled approach to method selection, based on theoretical analysis and empirical benchmarking to link characteristics of the problem with the suitability of the method.

ii. Standardization of validation procedures, such as benchmark data, test repeatability, and report checklists in MCDM studies.

iii. Dynamic and adaptive MCDM frameworks—where preferences, criteria and alternatives change over time, which are formally modelled and guaranteed in a theoretic sense.

iv. Efficient and transparent group decision making mechanisms that can be scaled up to large and heterogeneous groups of stakeholders with conflict resolution.

v. Explainable AI-MCDM integration ensuring interpretability, trust and ethical alignment in AI-augmented decision support.

vi. Human-AI Collaborative MCDM Paradigms based on the complementary strengths of human judgment and computational analysis.

vii. Reusability of domain-specific MCDM ontologies comprising reusable criteria, methods and validation practices for major application areas

The MCDM field is at a turning point: the fusion of traditional operations research rigor and artificial intelligence scalability. Sustained methodological innovation, built upon the principles of decision theory and proven through rigorous, reproducible practice is needed to realize the potential of this convergence.