Evaluation, classification, ranking, and related tasks should reflect human-observable properties of reasoning. The same principle should apply when machine learning models learn aggregation parameters, incorporating a human-in-the-loop to guide aggregation behavior. The goal of the proposed special session is to explore the latest theoretical advancements in logic-based data aggregation approaches and their applications in the field of data science. As data science becomes increasingly complex, the integration of logic-based methods becomes essential to develop robust models capable of addressing real-world challenges. Classification and evaluation should not neglect complex statistical and logic relations between attributes and their values, while solutions are explainable and interpretable. We are particularly interested in both novel developments and recent advancements in well-established logic-based approaches, such as the Choquet integral, Ordered Weighted Averaging (OWA) operators, ordinal sums of logic functions, logic aggregation based on the Interpolation-Based Approach (IBA), and Logic Scoring of Preference (LSP), among others.
Finally, this special session will provide a forum for researchers and practitioners to share progress and applications of logic-based data aggregation techniques across diverse domains, including ethics, business and finance, healthcare, computer vision, smart cities, ecology, and beyond.