Malzeme Biliminde Dijital İş Akışları ile Yapay Zeka Hazırlığının Desteklenmesi

Özet (abstract)

Materials science produces heterogeneous data across experiments, simulations, and industrial processes that often remain bound to local formats, manual procedures, and project‐specific software. Digital workflows address this fragmentation through explicit, repeatable, and machine‐actionable pipelines. This article examines 13 workflow contributions from the second and third funding phases of the MaterialDigital initiative. The contributions cover data acquisition and FAIR storage, simulation automation, multiscale integration, and AI/ML‐driven optimization. Here, AI‐readiness denotes the documented capacity of workflows and their artifacts to be reliably interpreted, executed, assessed, reused, and, where intended, invoked or adapted by automated systems; it does not require the direct application of artificial intelligence. Eight comparison aspects capture data and metadata, reusable artifacts, orchestration, robustness, cross‐scale coupling, transfer validation, learning and optimization, and adaptive or agent‐accessible operation. By distinguishing demonstrated capabilities from plans, the comparison shows how workflows support AI‐ready research both through integrated AI methods and through structured, traceable, and reusable pipelines. Across the projects, stable data structures, persistent artifacts, executable orchestration, and scientific validation recur as foundations, whereas adaptive operation and agent‐accessible interfaces remain specialized. Wider interoperability would additionally benefit from shared interfaces and explicit, portable descriptions of workflow artifacts.

Yazarların özeti; kaynağından alınmıştır. Advanced Engineering Materials, 2026 · DOI ↗