Topology optimization for smart and architected materials: A systematic review and semantic synthesis via neural topic modeling

This study develops a systematic–semantic review to map and synthesize the fast-growing intersection of topology optimization (TO) with smart and architected materials. Following PRISMA-based identification and multi-stage screening, a curated corpus was constructed to reduce semantic noise and domain drift and to ensure methodological relevance. A hybrid neural topic modeling pipeline was then implemented by coupling rigorous text normalization with transformer-based semantic embeddings and nonlinear structure discovery, enabling latent themes to emerge from semantic regularities rather than pre-imposed categories. The descriptive landscape indicates an inflection after 2021, consistent with the field’s maturation toward function-oriented material–structure co-design. Topic modeling reveals semantically coherent trajectories that integrate optimization methodologies, material functionalities, and computational strategies; keyword profiles and global semantic emphasis indicate a convergent shift toward learning-assisted design, multi-physics/multi-scale formulations, and fabrication-aware TO. Thus, the proposed framework provides a reproducible, high-resolution evidence base for identifying emerging trends, structural knowledge gaps, and actionable research directions in TO-enabled smart material systems.