Dr James Hester1
1ANSTO, Sydney, Australia
We aspire to a world where data described using differing standards can be ingested automatically based on machine-readable data description files: interoperability. A variety of machine-readable ontological languages for these data description files exist. Given this variety, building cross-community adoption of a single ontological standard is unlikely to be successful when integrating data from many domains, not least because fixing on a particular standard is likely to invalidate potentially decades of work developing and implementing otherwise solid intra-community standards in one or more target domains. This presentation explores one way to resolve this impasse: by making the ontologies interoperable, that is, by automatic transformations of the ontological descriptions themselves.
The relational model provides a universal basis for data description, in that the contents of any data file can be described as a set of tables. The ontologies that describe data and metadata from a given domain are simply “data about the data description” and also have a tabular representation. Data dictionaries expressed in one machine-readable vocabulary and format can therefore be machine-transformed into data dictionaries expressed in a different vocabulary and format using a suitable relationally-aware language.
This presentation describes how a data dictionary for images written in the DDL2 ontological language is algorithmically transformed to a dictionary written in the newer DDLm language. The compact, relationally aware dREL language is used to specify the transformations, which are performed after translating from dREL to Julia.
Dr James Hester has worked for over 20 years at both synchrotron and neutron facilities as an instrument scientist. At the same time as being involved in producing copious quantities of data, he has worked extensively on developing the next generation of IUCr CIF data standards, and is the current chair of the IUCr committee for the maintenance of the CIF standards.