[9p-W321-1~14] 23.1 Joint Session N "Informatics"

[9p-W321-3] Materials-dictionary set construction for Materials informatics

〇Akira Suzuki¹, Eiki Takayama¹, Masashi Ishii¹ (1.NIMS)

Keywords:Materials Informatics, text data mining, named entity recognition

Efficient collection of training data has become an issue for Materials Informatics (MI), and text data mining from enormous numbers of scientific articles is one of promising methods. We have studied automatic construction procedure of materials-dictionary set for data mining. Technical terms such as units, physical property names, sample names, measurement conditions were extracted using explored rules and archived with related texts into relational database (RDB) system. These data were improved by named entity recognition (NER) analysis combined with machine learning techniques.