Abstract Artificial Intelligence (AI) has become increasingly important in critical domains such as medicine, where accurate and interpretable decision-making is essential. However, many high-performing AI models operate as “black boxes”, limiting transparency and making it difficult for clinicians to understand or verify predictions. To address this challenge, we present an eXplainable Artificial Intelligence (XAI) framework that integrates a fuzzy rule-based classifier with genetic algorithms and 2-tuple linguistic representations. The method incrementally generates general fuzzy rules, introduces fuzzy exception rules to capture atypical cases, and applies rule selection and parameter tuning to enhance both accuracy and interpretability. Experiments on nine medical datasets demonstrate that our approach achieves competitive or superior accuracy compared to state-of-the-art algorithms, while requiring fewer rules. These results show that the method not only improves predictive performance but also provides clear, human-readable explanations for each decision, thereby increasing trust and facilitating its application in medical practice.

In Granular Computing, a key approach has been the handling of incomplete fuzzy preference relations through an allocation of information granularity and its corresponding optimization procedure. This methodology enables the transformation of numerical models into granular versions, offering a more accurate representation of reality by recovering missing information. In decision-making contexts involving fuzzy preference relations, it has played a crucial role in advancing procedures for estimating incomplete information. However, although several granularity allocation protocols have been proposed, only one has been employed so far: the one based on a uniform and symmetric allocation of information granularity. To address this gap, we aim to determine the efficacy of existing protocols for allocating information granularity in estimating missing values of incomplete fuzzy preference relations. Numerical tests are presented to demonstrate the effectiveness of each protocol.

Due to their unique structure and powerful capability to handle uncertainty and partial reliability of information, Z-numbers have achieved significant success in various fields. Zadeh previously asserted that a Z-number can be regarded as a summary of probability distributions. Researchers have proposed various methods for determining the underlying probability distributions from a given Z-number. Conversely, can a Z-number be used to summarize a set of probability distributions? This problem remains unexplored. In this article, we propose a nonlinear model, termed Maximum Expected Minimum Entropy (MEME), for generating a Z-number from a set of probability distributions. Through this model, Z-numbers can be generated directly from data without requiring expert knowledge. Additionally, we applied the MEME model to classification problems, introducing a novel if-then rule form, termed Z-valuation if-then rules. These rules replace the deterministic consequent part of a fuzzy rule with an uncertain Z-valuation, thereby further summarizing the uncertain information in the rule's consequent. Based on the Z-valuation rules, we propose a Z-valuation rule-based (ZVRB) classification system, which aims to enhance decision-making processes in scenarios where uncertainty plays a key role. To validate the effectiveness of the ZVRB classification system, we conducted two experiments comparing it with both classic and advanced nonfuzzy classifiers as well as fuzzy classification systems. The results show that the ZVRB model is superior to the other comparative classifiers in terms of classification performance.

Large-Scale Group Decision-Making, propelled by the advent of Large Language Models and the imperative for assertiveness in decision-making processes, emerges as a pivotal area of research. This paper navigates through the landscape of Large-Scale Group Decision-Making, delineating its significance across diverse domains, including social networks and e-democracy. Amidst its nascent status, Large-Scale Group Decision-Making encounters formidable challenges, particularly in information management and fostering consensus among a multitude of experts. This paper aims to illuminate the goals and hurdles facing Large-Scale Group Decision-Making approaches through an exhaustive review of contemporary literature and methodologies. By confronting these challenges head-on, Large-Scale Group Decision-Making holds the potential to redefine decision-making paradigms, bolster assertiveness, and elevate collective problem-solving capabilities in an increasingly interconnected world.

Group decision-making is a daily process where a group of experts has to choose from a set of alternatives. To help this group of experts, methods were developed that allowed them to recommend an alternative or a group of alternatives from the information generated by the experts, these methods were called Group Decision-Making methods. Nonetheless, some Group Decision-Making methods have some problems, such as evaluating alternatives in a general way. This problem leads to a loss of information because it is not possible to evaluate each alternative in detail and it may be the case that one alternative is better than another on one criterion but not on another. Another problem is that experts have priority for each criterion and for one expert one criterion may be more important than another and vice versa. To solve these two problems, in this novel method, we propose a consensual multi-criteria Group Decision-Making system, where experts can customise the order of the criteria. With the order of each expert, the method creates an aggregation operator for each criterion. Moreover, this method performs a consensus analysis for each criterion, so that there is a minimum consensus on each criterion. In this way, a solution is obtained that allows the set of alternatives to be evaluated in a detailed way and each criterion has the correct importance because the experts are the ones who create the order of importance.

Debate is a process consisting in arriving at a reasoned opinion on a proportion in which individuals must be truly capable of defending their own judgments. It has been used within group decision-making (GDM) problems to help experts make better decisions. However, whether experts engage in a vigorous debate, it can result in the use of aggressive language that may diminish consensus, which is the major objective of GDM. To avoid it, we present a novel method for GDM problems that can identify aggressive comments during the debate by incorporating a classifier based on sentiment analysis techniques. According to the information extracted during the debate, two procedures are developed to assign weights to the experts, which are used to introduce two new consensus measures and to make the final decision. Unlike the existing GDM methods, this new one can take advantage of the information extracted during the debate (i.e., experts’ behavior) throughout the decision process, making it in rapport with real-world GDM processes.

A group decision-making process is considered which is meant as that a group of experts (agents, decision-makers,…) rank a finite set of options from the best to the worst. A special class of such processes is discussed in which the number of experts is large or indeterminate, the so-called Large-Scale Group Decision-Making. In this type of process, a key factor is trust in making a decision and evaluating an alternative, and the problem of managing the trust of agents is in this type of process complex and challenging. In this paper, a new approach to the management of trust in a Large-Scale Group Decision-Making system is presented. For this purpose, clusters are formed based on two factors: the mutual trust that agents have for each other and the similarity of opinions. If these two conditions are met, the experts are grouped into a single cluster. In this way, it is possible to manage the trust of experts and to apply it to the formulation and solution of a Large-Scale Group Decision-Making system. It is also possible to detect isolated points, which are clusters of single experts.

Fuzzy Decision Making is a line of research that has had an important presence in the literature since 1965 until nowadays. From that year to the present, the line of research has been consolidated. The research field goals have evolved along with all the research that makes use of these systems and the new technologies requirements. Therefore, new challenges, different from the ones that were faced at the beginning, need to be solved. The aim of this paper is to describe the recent challenges faced by those working in this line of research. This paper will help researchers to know how their work should be oriented in the future.

When developing a Machine Learning model, the consideration of explainability as an additional design driver can improve its deployment into any application context. Given an audience, an explainable Artificial Intelligence system is one that produces details or reasons to make it's functioning clear or easy to understand. Among different paradigms that inherently support these capabilities, Fuzzy Rule Based Systems are a very accountable solution. The main issue when dealing with fuzzy systems is to select an appropriate granularity to represent (fuzzify) the input data. A low value may cause the generation of too generalist rules, causing a hinder on predictive performance, whereas a high value may lead to both overfitting and/or very complex solutions. To overcome this situation, we propose a novel hierarchical fuzzy classification system based on fuzzy exception rules. To do so, low granularity rules are first generated and their confidence is examined. For those cases in which the fuzzy confidence is below a quality threshold, new higher granularity rules are created to cover the "instances in conflict" for the general rule, which is still kept in the rule base. Experimental results show the achievement of a compact and interpretable final rule base while maintaining or improving the predictive performance in comparison with the baseline fuzzy rule based classification and hierarchical systems.