Meeting Notes 12.03.2021

Participants:

• Casper Andersen
• Rickard Armiento
• Francesca Bleken
• Stuart Chalk
• Jesper Friis
• Saulius Grazulis
• James Hester

Ontologizing CIF dictionaries

The main topic of discussion centered around the latest iteration of how to describe/ontologize the CIF dictionaries. See issue #16 to follow the progress.

Initial notes

These rather self-contained sentences were taken as part of quick sub-points that were made during the initial part of the meeting.

Attach attributes to data names. Most important is type.
Make CATEGORY a property of a data item.
Something is missing between SPACE_GROUP_SYMOP and the _space_group_symop_*.
CATEGORY can only be retrieved from the dictionary, not the CIF file itself.
Saulius auto-generated some ontology from the DDL dictionary.

Questions and topical discussion

Should CATEGORY be taken as a super-class of its data items or part of defining what a data item is?
We want to keep a notion of the hierarchy of data keys/items, e.g., _space_group_symop_id being a sub-class of a space_group_symop category. We don’t want to have a single level of the hierarchy with all data keys/items under DATA_ITEM, since we then lose some conceptual sub-sets of the data keys/items. But is this possible?

We’ve decided to add a super-set of the more specific data keys/items, e.g., the specific _space_group_symop_id isA subclass of _space_group_symop_, which hasCategory exactly 1 SPACE_GROUP_SYMOP and isA DATA_ITEM, while SPACE_GROUP_SYMOP isA CATEGORY and DATA_ITEM hasSpatialDirectPart CATEGORY. This effectively splits CATEGORY and DATA_ITEM taxonomically/in the class hierarchy, while still defining the sub-sets of data keys/items.

Practically, split up the CIF ontology/-ies into a “basis” or “DDL” ontology that contains everything down to DATA_ITEM and the individual _types. Then create “core” from the core CIF dictionary, e.g., SPACE_GROUP_SYMOP, etc.