A knowledge-driven fuzzy logic framework for supporting decision-making entities

Abstract

Decision support systems enable decision makers (whether individuals, systems, or other agents) to select the most suitable options by integrating expert knowledge with computational intelligence. Accurate modeling of these decision makers is crucial to ensure optimal decision making in complex and uncertain environments. Embedding expert knowledge in these models is challenging, as experts often lack familiarity with the underlying techniques. Therefore, there is a need for frameworks that are intuitive for experts and enable them to seamlessly integrate their knowledge into decision support systems. This paper presents a novel framework for the automatic generation of fuzzy decision models based on expert knowledge, designed to support decision-making scenarios. The proposed approach leverages the Takagi–Sugeno–Kang Fuzzy Inference System (TSK FIS) to model qualitative human reasoning and automatically induce decision models through expert-defined parameters that model the expert knowledge. This framework represents decision variables using linguistic terms, and introduces a weighted co-occurrence mechanism that captures variable interactions, enabling the generation of cumulative fuzzy decision rules that produce robust and interpretable outcomes. It simplifies expert data input through an intuitive method for defining relationships between variables, eliminating the need for extensive knowledge of fuzzy logic. The flexibility of the proposed framework is demonstrated through two practical case studies: passenger train ticket selection, and weapon choice optimization in video games, showcasing its effectiveness across varied domains. Experimental results validate the system’s capacity to generate tailored decision models that adapt to specific user profiles and objectives, while maintaining both decision-making accuracy and interpretability.