Fuzzy Rule-Based Approach for Weighting Artificial Experts Involved in a Multi-criteria Group Decision-Making Problem

Multi‐Criteria Group Decision‐Making (MCGDM) systems extend the techniques of Multi‐Criteria Decision Making (MCDM) to situations in which multiple decision-makers evaluate alternatives to produce an aggregated alternative. MCDM techniques handle complex problems by identifying alternatives and criteria and then assigning weights representing the relative importance of each criterion [22]. When multiple decision-makers evaluate alternatives, experts may differ in knowledge, goals, and biases, and equal treatment can distort outcomes [1]. Researchers have classified expert weighting methods into subjective techniques, where a supervisor assigns weights or experts evaluate themselves, and objective methods that use mathematical approaches such as distance-based or consistency‑based metrics to derive decision-maker weights [1]. This task has the potential to be particularly challenging since human experts tend to be averse to being assigned a weight. In light of these insights, we propose substituting human experts (i.e., decision-makers) with artificial experts in Fuzzy MCGDM systems, either partially or completely. In this framework, algorithms act as virtual experts whose judgements can replace or complement those of human experts. The methodology involves defining criteria and alternatives as in Fuzzy MCGDM, but the experts' weights are determined using a Fuzzy Rule-Based System. This involves defining a set of ad hoc rules to determine the fuzzy weights of the artificial experts. Furthermore, a supervisor (i.e., a human expert) could establish or update these rules. The integration of artificial experts is expected to mitigate human biases and enhance consistency. The proposed methodology aims to advance group decision-making towards more transparent, efficient, and evidence-based outcomes by embedding artificial decision-makers into the Fuzzy MCGDM process and using formal weight assignment methods. The article will detail the steps required to implement this methodology and discuss evaluation criteria for algorithmic experts.