Future of Adapter/RDF

[MBueschelberger]

Dear BaSyx-Developers,

We are currently using quite frequently the Experimental/Adapter/RDF for exporting AAS into RDF.

This is especially useful for us when we are transforming knowledge graphs in RDF following the PMD-Core ontology into AAS instances in JSON and vice versa in several of our use cases.

Please see our minimal demonstrator for this on GitHub.

Since this is implemented on a side branch, we see a potential risk of incompatibility issues with ongoing main branch developments or independent updates to the AAS meta model schema.

Therefore, we would like to emphasize that the future maintenance of this particular adapter is of great interest to us. Additionally, RDF deserialization of AAS through the BaSyx SDK would be highly valuable for our work, as demonstrated in our KG2AAS demo.

Do you already have come to a conclusion how you will handle the future developments of this feature?

Best regards

[s-heppner]

Hi, thank you for reaching out!

As mentioned in this comment and the following discussion, our official stance on the RDF serialization and deserialization can be found here.

To summarize, there are fundamental flaws with how the specification calls for reading and writing AAS to RDF.
Since we cannot have a fundamentally broken feature in our main branch, we kept it on its own "experimental" branch, which, as you may have noticed, we (as in the core developers of this project) do not actively maintain. We're happy to take Pull Requests though, if someone were to update it.

So to answer your question:

Do you already have come to a conclusion how you will handle the future developments of this feature?

To my knowledge, the specification issues with RDF have not been solved by the IDTA to this date, please correct me if I am wrong.
Until this is the case, the Eclipse BaSyx Python SDK will not officially support RDF.

[mhrimaz]

@MBueschelberger
I am glad that you are using the RDF representation of AAS.
As mentioned there are many different issues and they are already highlighted in the aas-specs-metamodel repository.

I did an implementation for my master thesis a while ago (before basyx-python-sdk work on their new release) you can explore that https://github.com/mhrimaz/py-aas-rdf. It fixes the issue that we have for ordering of keys by introducing aas:index attribute. This is necessary to perform the backward conversion from RDF to JSON. However, that tool is a shortcut for now and handling RDF should happen in official SDKs like aas-core-works or aas4j.

The IDTA Ontology Workgroup has been working on addressing the many issues mentioned. We are close to finalize the changes. In parallel, I am implementing these changes in the experimental branch of my py-aas-rdf repository. You can take a look at it or join the IDTA Ontology Workgroup if you are interested (your organization must be an IDTA member).

[MBueschelberger]

Dear @mhrimaz,

Thanks a lot for sharing this library! This package is exactly what we were looking for since a couple of months and it is exactly solving the problem which we partially wanted to tackle in the demonstrator. It works quite well with the use case we are addressing in the repository shown above. It is also nice to see that it is based on pydantic :)

Please see this script where we are applying your package for the Knowledge graph to AAS-transformation.

For stability reasons of the demo, we would fork the repository for py-aas-rdf for now.

However, we would be very happy if py-aas-rdf would become an official part of the official SDKs!