The Starrydata web system streamlines manual data collection by efficiently extracting experimental data from figures and tables in scientific papers, enabling the construction of an open-access materials database. In this study, conducted in collaboration with Murata Manufacturing Co., Ltd., approximately 20,000 temperature-dependent datasets on perovskite-type oxide dielectric materials were collected. Machine learning was applied to predict material properties, identify key compositional factors, and visualize trends across existing studies, demonstrating the potential of data-driven approaches in materials research.

Created at: 2025-06-11 03:07:13 +0900 Updated at: 2026-03-21 04:34:00 +0900