Technologists from the National Institute for Materials Science, Japan, working in collaboration with the Toyota Technological Institute, have built a Computer-Aided Material Design system that can extract data pertaining to fabrication processes and material structures, together with material properties. Each of these aspects are critical to material design.
The scientists have then used the outputs to organize and visualize the relationship between materials. This process enables vast quantities of information, drawn from thousands of scientific articles, to be summarized in a single chart and for this chart to be used to help expedite material design.
Computer-Aided Material Design is based on the premise that producing useful new materials could become more straightforward provided their properties can be calculated prior to producing them. In most cases, a complete description from first principles is too complex; however, computer-aided materials design provides an enabling progress by linking the electronic, atomistic, microstructural and continuum levels, resulting in efficient design practices.
This is based on materials informatics, which is an information science-based approach designed to assist with materials research. This enables the relationships between these factors to be extracted from large amounts of data using deep learning processes. The use of artificial intelligence helps with the sorting of the data and with the review and detection of patterns.
In trials, the Japanese researchers have produced a system that can extract and identify relationships between factors lined to processes, structures and properties required for material design. They achieved this by using an artificial intelligence to read the text of scientific articles via natural language processing.
Through this process the artificial intelligence extracted the relevant information and determined the type of relationships between material structures necessary to achieve desirable material properties. An example used was with strengthening stainless steel.
The researchers have provided the artificial intelligence source code for use by others for free, provided that the code is use for related research.
The research has been published in the journal Science and Technology of Advanced Materials, with the research paper titled “Relation extraction with weakly supervised learning based on process-structure-property-performance reciprocity.”