2023年度 DxMT人材育成セミナー (第3回)を、2023年6月28日に開催しました。
参加者数: 145名
【講師】
北海道大学大学院 理学研究院
高橋 ローレン 特任助教
【タイトル】
"Ontology and the road towards knowledge-driven material design"
【セミナー内容】
Ontology is increasingly becoming an important component of data organization and utilization within the materials science research community. The exponential growth of data availability and improved accessibility has allowed researchers to adopt data science as a fourth approach to material research, leading towards the development of materials informatics. However, researchers are faced with many issues regarding data and databases such as incompleteness, biases, and ambiguity- all which inhibit proper informatics development and implementation. Additionally, the ability to model scientists' observations and experiences in a way that is accessible by machines has become necessary in order to develop proper knowledge-based frameworks for machine-driven material design.
Ontology is proposed as an alternative method for curating and managing material data. With its origins in philosophy, ontology provides the ability to incorporate meaning and relational information with raw data, thus introducing a more interconnected and semantics-driven method of data interaction. (J. Chem. Inf. Model. 2018, 58, 9, 1742–1754) Applications of ontology towards several databases and the periodic table and their benefits towards data merging and querying are also explored.
Adopting ontologies allows for knowledge-based data querying while also improving materials data standardization and materials classification where proper development can lead towards a true implementation of AI towards materials and catalysts design. Ontology also provides a way to define researchers’ experiences and knowledge with scientific research in a way that preserves data relationships and meanings while remaining accessible by machines. (J. Phys. Chem. Lett. 10 (23), 7482-7491) With its ability to merge and query databases together in a single space, it has become possible to use ontology to create a knowledge-based framework for machines, thus taking a step towards developing a true knowledge-based AI.