A FUZZY DECISION SUPPORT SYSTEM FOR EXPLAINABLE EMOTION ANALYSIS

Abstract

The goal of the task is to develop a system, corresponding to the concepts of fuzzy logic, to objectively analyze the user's emotions for decision making. It is proposed to solve this problem through Explainable Artificial Intelligence (XAI) using Fuzzy OWL2 Ontology. The ontology of emotional states is based on predefined fuzzy concepts: 'stress', 'anxiety', 'moderation', 'joy' modeled in Fuzzy OWL2. Decision ontologies are fuzzy reasoning rules linking emotional states to decision outcomes (e.g., “If stressed, decision accuracy is lower”). The contextual data annotations are a mapping of contextual aspects of the experiment (time, task type, individual traits) using the ontology. The OWL2 fuzzy ontology allows combining definable data types into a semantic reasoning model that consists of collecting subjective data in real time, mapping the raw data to fuzzy linguistic categories such as “low”, “moderate”, or “high”, and using fuzzy reasoning to derive emotional states and decision models based on the ontology. The result is a generated understanding of emotional states and decision-making quality in fuzzy conditions. The system must cope with categorizing emotional states and relate them to decision making in fuzzy conditions, with fuzzy logic and XAI improving transparency and interpretability. The uncertainty of the data, the need for physical data that is not provided in this experiment, the variability of human responses and the complexity of the system limit the actual results. Expected result – emotional state classification accuracy ≥85%. Actual result – accuracy varies (70-90%) depending on data quality and feature diversity. Improved feature development, robust ontology design and more accessible XAI visualization tools are required that can reduce the difference between expected and actual results.