Skip to content

Meeting notes 21.05.2021

Jump to bottom
Casper Welzel Andersen edited this page Jun 18, 2021 · 2 revisions

Participants:

  • Jesper Friis
  • Rickard Armiento
  • Casper Andersen (sort of)
  • Emanuele Ghedini

Deeper analysis of CIF ontologization

Casper was driving, so it turned out mostly to be a discussion between Rickard, Emanuele, and Jesper about the consequences (or not) on the CIF ontology from the deeper analysis Emanuele has been working on by ontologizing essentially the full CIF specification.

After some general discussion centered mostly on Emanuele's latest adjustments of the meta-CIF-ontology, Rickard tried to bring the discussion back to what he had understood our original plan was:

  1. Create an ontology to represent the semantics and the contents of CIF files (but not their syntax).
  2. Create a more EMMO-integrated crystallography ontology. This should probably be based on what we've learned from (1), but can be free from the CIF semantics and adopt EMMO concepts, etc.
  3. Create the relationship between individuals and concepts in (1) and (2), so we can map between these and reason on them.

This led to two things:

  • Casper thinks we need to get the ball rolling on (2), and - after some back and forth - we probably should do so by moderating our discussions going forward so they cover both (1) and (2) and we don't get stuck on too much on the details of (1).
  • Emanuele, Jesper, and Rickard dove deep into the discussion about data items and categories in CIF - because this can really be seen as a representative discussion about something in our present representation of (1) that Emanuele disagrees with.

The end conclusion about the latter point was essentially that Emanuele should come up with concrete suggestions about how the diagram that we presently have for (1) needs to be adjusted, along with the arguments why, and we should take that discussion with Saulius and James present (since they are the CIF experts and the right people to comment on whether these modifications agrees/disagrees with their own models of CIF).

Category concept in CIF top

Emanuele feels strongly that we need to remove/reconsider Category from the CIF ontology with arguments deriving from the theoretical basis of ontologies.

The main point of contention is the understanding of Category, and whether it is to be understood in a more meta-ontological manner, being quite abstracted or as another label used in the CIF ontology that represents a set of terms. In essence, Emanuele argues for the first point, since what we are trying to do is ontologize the structure and definition layout of CIF, where Category is not only a label for a set, but represents a set of sets, adding an extra ontological dimension that we cannot define within the current scope. Rickard argued for the latter, that Category is merely a label to represent the different CIF categories laid out in the dictionaries. Each of these CIF categories, e.g., SPACE_GROUP_SYMOP, contains each of that category's fields, e.g., _space_group_symop.id.

As a reference, Rickard mentioned the documentation page: https://www.iucr.org/__data/iucr/cifdic_html/1/cif_ms.dic/Cspace_group_symop.html, where it is stated "The SPACE_GROUP_SYMOP category that introduced [...]", and asked Emanuele whether he would then agree that SPACE_GROUP_SYMOP isA Category, but Emanuele argued that this is a colloquialism and does not support the semantic notion that SPACE_GROUP_SYMOP isA Category, but merely that SPACE_GROUP_SYMOP isA Categorized (or something similar).

Since we were missing the CIF experts to shed more light on the IUCr definition and the referenced description, the discussion ended here and will be taken up again at a later point.

Clone this wiki locally