Initial version of PyCRAM Designators in SOMA by mrnolte · Pull Request #320 · ease-crc/soma

mrnolte · 2025-05-12T12:29:13Z

Here is a minimal addition of PyCRAM Designators to SOMA via SOMA-PYCRAM. Does it make sense? And, especially, does it make sense to put it into the main repository?

update

mpomarlan · 2025-05-12T12:48:56Z

owl/catalog-v001.xml

-	<uri id="User Edited Redirect" name="http://www.ease-crc.org/ont/SOMA-AGENT.owl" uri="SOMA-AGENT.owl"/>
-    <uri id="User Edited Redirect" name="http://www.ease-crc.org/ont/SOMA-ALL.owl" uri="SOMA-ALL.owl"/>
+    <uri id="User Edited Redirect" name="http://www.ease-crc.org/ont/SOMA-AGENT.owl" uri="SOMA-AGENT.owl"/>
+    <uri id="User Edited Redirect" name="http://www.ease-crc.org/ont/SOMA-All.owl" uri="SOMA-ALL.owl"/>


SOMA-ALL became SOMA-All in the name. Is this safe or will it break anything?

This was because the prefix in the file SOMA-ALL was actually SOMA-All, I fixed this

mpomarlan · 2025-05-12T12:50:26Z

owl/SOMA-PYCRAM.owl

@@ -0,0 +1,66 @@
+Prefix(:=<http://www.semanticweb.org/robin/ontologies/2025/4/untitled-ontology-8/>)


A minor nitpick but the : prefix should be something SOMA related, rather than a Protege generated-default.

Fixed this!

mpomarlan · 2025-05-12T12:54:26Z

owl/SOMA-PYCRAM.owl

+
+# Class: <http://www.ease-crc.org/ont/SOMA.owl#PyCramActionDesignator> (PyCRAM Action Designator)
+
+AnnotationAssertion(rdfs:comment <http://www.ease-crc.org/ont/SOMA.owl#PyCramActionDesignator> "A PyCRAM Designator expressing a Plan for a Task.")


AFAIK action designators are not exactly plans. They specify what, not how.

If possible, I would go with any "official" definitions for pycram action designators here. E.g., what does the pycram documentation describe them as? If it turns out the pycram doc describes them as plans, it is itself likely wrong however and we would need a different text for them.

Well, I'd argue that DUL does not offer anything better than a DUL:Plan for this, unfortunately. I quote the documentation (https://pycram.readthedocs.io/en/latest/designators.html):

Action designators can be seen as higher level plans that include failure handling and parametrization.

Ontologically, they are of course not the plan itself, merely a structured representation of said plan; i.e., they must be a DUL:InformationObject (more specifically, they are SOMA:StructuredText, but this does not matter for this argument). As a DUL:InformationObject, they DUL:expresses some DUL:SocialObject. This is true for all PyCRAM Designators (!). Now, for Action Designators we may consider different SocialObjects, e.g., DUL:Task, DUL:Plan, and DUL:Method. For DUL:Task, we would need a DUL:Description that DUL:defines it. This to me is unintuitive, as while a DUL:Task may have subtasks, it is not structured - the structure stems from a plan (as far as I understand it). For a DUL:Plan, DUL says

A Plan is a Description of some actions to be executed by agents in a certain way, with certain parameters

This to me sound exactly to what Designators are. DUL:Method seems wrong to me, because it more vague, i.e., DUL says

It is different from a Plan, because plans could be carried out in order to follow a method, but a method can be followed by executing alternative plans.

So, when your saying that

If it turns out the pycram doc describes them as plans, it is itself likely wrong however and we would need a different text for them.

what do you actually mean by that? What do you think do they express?

mpomarlan · 2025-05-12T12:55:59Z

owl/SOMA-PYCRAM.owl

+AnnotationAssertion(rdfs:comment <http://www.ease-crc.org/ont/SOMA.owl#PyCramActionDesignator> "A PyCRAM Designator expressing a Plan for a Task.")
+AnnotationAssertion(rdfs:label <http://www.ease-crc.org/ont/SOMA.owl#PyCramActionDesignator> "PyCRAM Action Designator")
+SubClassOf(<http://www.ease-crc.org/ont/SOMA.owl#PyCramActionDesignator> <http://www.ease-crc.org/ont/SOMA.owl#PyCramDesignator>)
+SubClassOf(<http://www.ease-crc.org/ont/SOMA.owl#PyCramActionDesignator> ObjectSomeValuesFrom(<http://www.ontologydesignpatterns.org/ont/dul/DUL.owl#expresses> <http://www.ontologydesignpatterns.org/ont/dul/DUL.owl#Plan>))


See above. Worth discussing, because action designators, at least in Lisp CRAM, rather expressed goals than plans.

Ah I now see; however, we have that DUL:Plan DUL:hasComponent some DUL:Goal, i.e., we have something more general than the goal itself.

mpomarlan · 2025-05-12T12:59:09Z

owl/SOMA-PYCRAM.owl

+
+As structured data, it conveys:
+- For object lookups: a dul:Configuration of feature‐roles (e.g., type, color, shape).
+- For actions and motions: a dul:Method that prescribes how to carry out an underspecified task or movement toward a goal.


true for motion. As discussed above, at least this was not the intention for the action designators. The reason was that several different actions may be appropriate for achieving a goal, and thus the idea was to specify a common core -- the goal to be pursued -- and then let the designator resolution process to fill in the details of how the action is to be performed. Designators can add more constraints to this description, all the way down to fully specifying what actions to use and with what parameters, but this is not intended as a typical usecase.

Now, following your line of argumentation, I believe we must look at two different things here, actually. So, an unresolved action designator would express a DUL:Method, while a resolved action designator would express a DUL:Plan?

mpomarlan · 2025-05-12T13:04:59Z

owl/SOMA-PYCRAM.owl

+
+AnnotationAssertion(rdfs:label <http://www.ease-crc.org/ont/SOMA.owl#Resolving_of_PyCRAM_Designators> "Resolving a PyCRAM Designator")
+SubClassOf(<http://www.ease-crc.org/ont/SOMA.owl#Resolving_of_PyCRAM_Designators> <http://www.ease-crc.org/ont/SOMA.owl#DerivingInformation>)
+SubClassOf(<http://www.ease-crc.org/ont/SOMA.owl#Resolving_of_PyCRAM_Designators> ObjectSomeValuesFrom(<http://www.ease-crc.org/ont/SOMA.owl#isTaskOfInputRole> <http://www.ease-crc.org/ont/SOMA.owl#PyCramDesignator>))


What are the domains and ranges of the isTaskOfInput/OutputRole?

Ah damn, I forgot the Role inbetween. Will change that, good catch!

mrnolte · 2025-05-13T21:45:42Z

I now added more specific classes to the model, which divide the underspecified description (PyCRAM Designator Description) and the specified description (PyCRAM Designator Referent). @mpomarlan does this now resolve your issues with the modeling?

As a side note, there is yet another issue, in which the description itself, e.g., "a blue cup" must be somehow represented. This is problematic in that OWL does not allow nesting of expressions. I fear that we must reify the concepts and relations.

mpomarlan · 2025-05-14T11:48:03Z

Could you expand on the blue cup comment?

For the rest, it turns out we have 2, soon to be 3, PRs incoming about designators. Yours, Sascha's, and Alina's. It would be good I think to harmonize these in a SOMA meeting, e.g. this Friday.

mrnolte and others added 2 commits May 12, 2025 12:50

Merge pull request #32 from ease-crc/master

80d5e4d

update

update

d19f80d

mpomarlan self-assigned this May 12, 2025

mpomarlan reviewed May 12, 2025

View reviewed changes

mrnolte added 2 commits May 13, 2025 21:26

Fixed prefixes

0bd116e

Update SOMA-PYCRAM.owl

5ebc06c

mrnolte requested a review from mpomarlan May 13, 2025 21:45

mrnolte marked this pull request as draft May 13, 2025 21:45

mrnolte added 2 commits May 14, 2025 00:13

Fixed IRIs

d0d2643

fixed superflous axiom

2d0ae1c

mrnolte added 2 commits May 14, 2025 16:14

added is method for

ad024dc

Merge branch 'master' of https://github.com/mrnolte/soma

1721008

		@@ -0,0 +1,66 @@
		Prefix(:=<http://www.semanticweb.org/robin/ontologies/2025/4/untitled-ontology-8/>)


		# Class: <http://www.ease-crc.org/ont/SOMA.owl#PyCramActionDesignator> (PyCRAM Action Designator)

		AnnotationAssertion(rdfs:comment <http://www.ease-crc.org/ont/SOMA.owl#PyCramActionDesignator> "A PyCRAM Designator expressing a Plan for a Task.")

Conversation

mrnolte commented May 12, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mrnolte commented May 13, 2025

Uh oh!

mpomarlan commented May 14, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants