Potential overwriting of attributes in imported modules

[markus-rothkoetter]

Description of the issue

Changes introduced in #1086 (e.g. adding Inverse Of) reside in ro-module which is an imported module.
The content of the ro-module is automatically extracted via ROBOT 🤖.

Inspecting the extract-ro-module.sh-script, I have NOT seen any precaution checking for potential changes BEFORE recreating the ro-module from scratch/source.

From my perspective this means: As soon as the script is run, all previous edits are lost immediately.

Can someone comment whether this is true or if there any taken measures I'm not aware of?
@jannahastings, @stap-m, @MGlauer

Ideas of solution

In case I'm right about my assumption, I see the following options:

1. implement a diff check in the script and merge changes
2. change the location of implementation of attributes in imported modules

[jannahastings]

This is indeed true, and I think your option 1. is the better one. We did not so far re-run this script after initially creating the module, but there may be situations in the future where we would need to do so.

[markus-rothkoetter]

As this is the bottleneck for #1114 & #1023 (both require additional ro-relations), I created a simple diff/merging command in the script and created the ro-module from scratch. (mentioned as Option 1 before)

TL;DR: @stap-m @l-emele @jannahastings
The diff-check showed some issues (caused by past manual edits) which break a simple recreation of the file (see next comment)
With respect to the upcoming release:
I can merge the terms mentioned in #1023 manually via ROBOT into the existing ro-module and treat #1114 & #1023 afterwards. This means the script fix has to wait until the release is done.

Is this a valid approach?

[markus-rothkoetter]

---

BFO_0000019 aka quality resides as a class in ro-module
(This is introduced by the annotation property is homeomorphic for)
🔑-question no. 1: Shall this be moved to the defining-ontology of quality?

BFO_0000040 aka material entity is pulled into the module by RO_0002577 aka system.
I found the commit where @jannahastings added this to the ontology and another commit removing major parts of BFO_0000040 class definition. (I assume in order to remove duplication?)
🔑-question no. 2: I assume this shall remain, so I'm going to add this to the extract-no-hierarchy.txt?

RO_0000086 aka has quality is in the ontology (including subclasses), but not in the extract-lists for the script.
Added during commit
🔑-question no. 3: Shall I add this to the extract.txt?

IAO_0000115 aka definition, an annotation property, resides in ro-module
🔑-question no. 4: I think it belongs into the iao-module. Shall I move it there?

oboInOwl:inSubset is now added to multiple properties, when recreating the ro-module
🔑-question no. 5 : @jannahastings Is this needed or overhead? Shall we excluded it via ROBOT?

foaf:page is an annotation property used once and there is also an import related to foaf.
I don't know what this does and how to treat it in the import script as the from-scratch version does not contain it anymore.
🔑-question no. 6: What to do? :)

Lastly, there are some definition updates different from the current oeo-imported-version e.g.:

• has output:
  "p has output ... and c is not present at the beginning of p."
  vs
  "p has output ... and c is not present in the same state at the beginning of p."
• has characteristic:
  "a relation between an independent continuant (the bearer) and a specifically dependent continuant (the dependent), in which the dependent specifically depends on the bearer for its existence"
  vs
  "Inverse of characteristic_of"

🔑-question no. 7: Shall the updated ro-definition be used or is there a different approach?

[jannahastings]

BFO_0000019 aka quality resides as a class in ro-module
(This is introduced by the annotation property is homeomorphic for)
🔑-question no. 1: Shall this be moved to the defining-ontology of quality?

Yes, `quality' should be a class in BFO. Isn't it already there?

[jannahastings]

BFO_0000040 aka material entity is pulled into the module by RO_0002577 aka system.
I found the commit where @jannahastings added this to the ontology and another commit removing major parts of BFO_0000040 class definition. (I assume in order to remove duplication?)
🔑-question no. 2: I assume this shall remain, so I'm going to add this to the extract-no-hierarchy.txt?

I don't remember why I did this but yes we don't want to duplicate any annotations so if we include the declaration it should be bare without additional axioms.

[jannahastings]

RO_0000086 aka has quality is in the ontology (including subclasses), but not in the extract-lists for the script.
Added during commit
🔑-question no. 3: Shall I add this to the extract.txt?

No objections, sounds sensible to me

[jannahastings]

IAO_0000115 aka definition, an annotation property, resides in ro-module
🔑-question no. 4: I think it belongs into the iao-module. Shall I move it there?

Yes (isn't it already there?) but in the medium term we need to move to using the 'obo metadata ontology' which is a new and updated version of the IAO ontology metadata tags such as these, which I think is the new primary source for commonly used shared annotation properties.

[jannahastings]

oboInOwl:inSubset is now added to multiple properties, when recreating the ro-module
🔑-question no. 5 : @jannahastings Is this needed or overhead? Shall we excluded it via ROBOT?

Indeed I don't think we need this. Although it might potentially be a good candidate for the annotation property which flags which module a particular class is defined in, see #870

[jannahastings]

foaf:page is an annotation property used once and there is also an import related to foaf.
I don't know what this does and how to treat it in the import script as the from-scratch version does not contain it anymore.
🔑-question no. 6: What to do? :)

Ugh can we remove this ?

[jannahastings]

🔑-question no. 7: Shall the updated ro-definition be used or is there a different approach?

The updates from RO look good to me.

[markus-rothkoetter]

No action required currently 🚧, just a note for our future updates to the script:
Ontology IRI and Version IRI are different in the current ro-module-version from the script-generated one.

script

 --ontology-iri http://open-energy-ontology.org/ontology/imports/ro-module.owl 
 --version-iri http://open-energy-ontology.org/ontology/imports/ro-module.owl 

vs

current module

 --ontology-iri http://openenergy-platform.org/ontology/oeo/imports/ro-module.owl 
 --version-iri http://openenergy-platform.org/ontology/oeo/dev/imports/ro-module.owl 

[markus-rothkoetter]

Discussions are moved to follow-up issue #1154
(This topic touches also other areas not covered in this issue.
So, information are collected in one place (#1154) for reworking the import process)