Enhancing Explainability and Accelerating Materials Science Design with Linguistic Summaries

Accelerating the discovery of advanced materials is a critical priority in materials science research. Multi-objective optimization frameworks like PyePAL have significantly streamlined this process by automating the search for Pareto-optimal solutions and guiding the sampling of promising design candidates. However, the interpretability of high-dimensional data and optimization outputs remains a pressing challenge, often limiting the practical impact of these methods on real-world applications. In this paper, we address this gap by integrating fuzzy linguistic summaries into the PyePAL workflow to enhance explainability and accelerate materials science design. We convert quantitative optimization data into intuitive, human-readable statements. This approach enables researchers to rapidly identify key trade-offs, discern patterns, and prioritize experimentation strategies without being overwhelmed by numerical details. We demonstrate our approach on a common multi-objective optimization problem to validate its effectiveness.