19th
International CODATA Conference
Category:
Interoperability
Web Ontology definition for design oriented material selection
Toshihiro Ashino (ashino@acm.org), TOYO Univ. Regional Development Studies, Japan, http://vaudesir.itakura.toyo.ac.jp/
Mitsutane Fujita (fujita.mitsutane@nims.go.jp), National Institute for Materials Science,
Several data schema definition of material properties for XML with XML Schema,
RelaxNG or other schema definition languages are
developed. They give common and exchangeable expression of material data. Next
stage toward knowledge management about material usage, selection or processing
is to define ontology which represents structure of concepts relate to material,
e.g., definition, classification or properties of material.
Defining ontology is a process to specify material class/subclass relationships, properties of each classes and restriction of properties. Basic structure of ontology class hierarchy can be derived from material thesaurus. But thesaurus has much less expressive power than ontology. It doesn't allow multiple inheritance, it can't define property restriction, class/instance relationships. The structure of material substance and properties are not distinguished in thesaurus definition.
Material selection for designing artifact is a process to derive required material properties from its utilization, to translate material definition specified in its property domain to materials in substance domain. This process requires ontology definition of appropriate domain. Web Ontology Language (OWL), be standardized recently as a part of Semantic Web, assures to create interoperable ontology definition and related software tools will become available.
Also, design process requires definition of rules to handle data, physical lows, empirical knowledge, but rule definition language is not standardized yet. We will show the ontology definition of creep properties as example, usage of it for design process and its limitations. The the role and applications of rule and logic, which include semi-empirical rules or data processing methods will be discussed.