[17p-A401-1~15] 23.1 Joint Session N "Informatics"

[17p-A401-12] Interpretation of transition temperatures in superconducting data extracted from literature

Kento Mitsui¹, 〇Ryoji Asahi¹ (1.Nagoya Univ.)

Keywords:materials Informatics, natural language processing, superconducting

Slot extraction of material composition, transition temperature, and process information was performed on 48,865 literature abstracts on superconductivity. Noting the variation in the transition temperatures obtained, density-based spatial clustering was performed for the material compositions, and then a Gaussian mixture model was used to classify the distribution of transition temperatures for each cluster. The Gaussian distribution of higher transition temperatures was interpreted as reflecting the researcher's trial and error in raising the transition temperature for that composition cluster.