Ontology development and machine learning for automating materials science research and materials discovery
Year: 2022
Type: conference-paper
Abstract: Industrial progress is strongly connected to materials development. New functional materials, specifically tailored to solve specific tasks, are often needed for enabling new technologies and industry processes and to develop new products. Materials science has a long history in enforcing high performance computing paradigms. Traditional approaches to computational materials science, however, are quickly reaching their limits both in throughput and in predictive power. New data-driven computational paradigms, including machine learning and deep learning, offer now opportunities both in replacing and supporting computational simulations. The widespread adoption of data-driven technologies in materials science is often spoiled by fragmentation and non-standardisation of materials data and computing approaches. Pairing the advances in data-driven computing technologies with efforts in knowledge engineering in the field of advanced materials can lead to substantial improvements, accelerating the development process and improving the quality of results. Here, we present some preliminary result in this direction, showing how the application of machine learning to materials science paired to a domain ontology specifically designed for molecular materials enabled us to accelerate the research on materials and materials discovery within a broad application domain.
Source: PORTO@Iris
Institutions: University of Bologna, Politecnico di Torino
Language: English
Subfield: Materials Chemistry
Field: Materials Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7219252147
