You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Due to the EMMO convention to name all concepts in the ontology with unique UUIDs instead of readable names, the source file become very difficult to read. That leads to unreadable diffs, making the review process very hard. It also makes writing SPARQL queries very difficult.
My advise is to drop the UUIDs and name the concepts after their prefLabel.
This is an ongoing discussion in the EMMO group. There will probably not be any changes in the near future.
We could decide to drop the UUIDs in the microstructure domain ontology. But that would not solve the problem, since all references (including relations) to EMMO top and middle would still be unreadable UUIDs.
Other alternatives:
Write the ontology in Python
Pros:
git friendly
easy to read source
Cons:
all changes to the ontology must be done in Python, creating an additional barrier for non-Python users to develop the ontology
the ontology cannot be distributed as-is on github, but has to be "compiled" into turtle for distribution (e.g. on GitHub Pages)
Add a pre-commit / post-pull hooks that always saves/reads the ontology to/from standard format
Pros:
possible to replace all UUIDs with readable names in the checked-out sources, given that we also checks out EMMO
possible to always store in a standard format avoiding false differences (makes the review process much easier)
Cons
does not solve unreadable diffs
complicated, may be error prone
an extra barrier to learn for contributors
Remarks
I don't think that the two alternatives mentioned above are worth the effort. It would be much easier and more explicit to simply replace the UUIDs with prefLabels. Note that this will not solve the issue of false diffs due to Protege serialising the ontology slightly different depending on OWLAPI version and how you save. But it will make the diffs much easier to read. It can further be minimised by standardising on OWLAPI version and how to save the ontology.
The text was updated successfully, but these errors were encountered:
Due to the EMMO convention to name all concepts in the ontology with unique UUIDs instead of readable names, the source file become very difficult to read. That leads to unreadable diffs, making the review process very hard. It also makes writing SPARQL queries very difficult.
My advise is to drop the UUIDs and name the concepts after their prefLabel.
This is an ongoing discussion in the EMMO group. There will probably not be any changes in the near future.
We could decide to drop the UUIDs in the microstructure domain ontology. But that would not solve the problem, since all references (including relations) to EMMO top and middle would still be unreadable UUIDs.
Other alternatives:
Write the ontology in Python
Add a pre-commit / post-pull hooks that always saves/reads the ontology to/from standard format
Remarks
I don't think that the two alternatives mentioned above are worth the effort. It would be much easier and more explicit to simply replace the UUIDs with prefLabels. Note that this will not solve the issue of false diffs due to Protege serialising the ontology slightly different depending on OWLAPI version and how you save. But it will make the diffs much easier to read. It can further be minimised by standardising on OWLAPI version and how to save the ontology.
The text was updated successfully, but these errors were encountered: