Ontologies and Databases – Knowledge Engineering for Materials Informatics

The discipline of informatics emerged from the need to translate biomedical research into evidence-based healthcare protocols and policy. Materials science informatics is rooted in an analogous need to “translate” physical sciences research and discoveries into materials-based solutions to address a broad range of issues and challenges for business, government, and the environment. Ontologies and databases are key elements of translational architectures and therefore are fundamental tools of the practice of informatics. Databases are tools for engineering data and information, while ontologies are tools for engineering knowledge and utility. Since knowledge and utility are the core objectives of informatics, correctly understanding and utilizing ontologies is critical to the development of effective materials informatics programs and tools. Rooted in philosophy, the term ontology appears most frequently today in connection with semantic web technology, where it refers to vocabularies used by inference engines to interpret human use of language. Materials science ontologies need to capture the scientific context of the defined concepts to support modeling and prediction of multidimensional structure–property relationships in variable environments and applications. Addressing the complexity of materials science ontologies requires a significant departure from traditional database and semantic web ontology approaches, including the use of neural networks that are capable of implementing methods for modeling context, relevance, complex systems, and human expertise. Pioneering efforts in this space include the Knowledge Engineering for Nanoinformatics Pilot (KENI) launched by the Nanoinformatics Society in 2010, and a collaborative Materials Genome Modeling Methodology initiative led by Iowa State University and initiated in 2011.