CIDOC CRM

Last week I attended a two day workshop at the Oxford eResearch Centre (OeRC) in order to get my head around semantic mapping of data using CIDOC-CRM. Run by Research Space, the training was delivered by Dominic Oldman, the Head of Research Space and Maria Theodoridou, an R&D Engineer at the Foundation for Research and Teaching, Helas.

The CIDOC conceptual reference model provides definitions and formal structure for describing the implicit and explicit concepts and relationships used in cultural heritage documentation.

The workshop started by illuminating what on earth this means! I’ll try and share my rudimentary understanding, but if you want to go deep, visit the CIDOC CRM website. As you will see from our first attempt at mapping pictured below against that of the professionals – I’m still working on it!

Our first attempt at mapping

Masterclass on mapping from the professionals

Basically, CIDOC-CRM transforms links between data from human readable format to machine readable format via the semantic web. For example, you have a database record for an object, with its unique identifier, material, production time and place, etc. Reading this database record it is easy for the human mind (or at least the curator’s mind) to parse how these pieces of information relate to one another. This information is not immediately obvious to a machine, and the relationships between the various bits of data have to be mapped.

CIDOC-CRM is one framework for doing that, taking an object based approach. Real world things are connected to other real world things through the use of properties or relationships. Dates, places, methods of making are all considered real world things, as are ‘events’ such as the production or acquisition of an object – objects are either physical and conceptual things, or temporal; in this way CIDOC-CRM aligns with object based programming languages. The connections and relationships are very general – ‘created by’, ‘has property’ in order to cover the multitude of possibilities.

The resulting data is extremely useful, not only because it is machine readable, but because it is the basis of linked data, allowing completely separate data sets to be cross-referenced. Individual objects are given unique identifiers. Further, universal objects such as people, places, dates and materials can be given universal unique identifiers. If mappers use these universal unique identifiers when mapping their data, the data then becomes cross-reference-able with other data-sets making use of the same standards. An example of this in practice is CLAROS, a linked data site for Greek and Roman artefacts.

Linking data using CIDOC-CRM

In the workshop we took real XML data records and practices mapping these, getting used to the highly structured rules that govern CIDOC-CRM. We looked at single records, but in the real world you would map an entire database structure, taking examples of records that utilise all the available database fields, and then testing that these work across the board.

Overall it was a great workshop and significantly increased my understanding of data mapping in general and CIDOC CRM specifically. If another course like this comes up, I would highly recommend it. It also included a nice dinner at Keble college…

SCR Dining at Keble College