Compliancy level with other patterns (e.g. SARI and Linked.Art patterns)

[VladimirAlexiev]

@stephenhart8 @chin-rcip: please consider the CH SARI model developed by @ncarboni and George Bruseker: https://docs.swissartresearch.net/et/persons/ Extra presentation of https://docs.swissartresearch.net/presentations/crmsig/.

Overall, this is a nice and sound model. I made some comments in tweets starting at https://twitter.com/wlpbloyd/status/1237775367148113921

[VladimirAlexiev]

The discussions have been moved to github: https://github.com/swiss-art-research-net/reference-data-models/issues which is a more appropriate medium

[illip]

Thank you @VladimirAlexiev, CHIN was already aware of SARI's work and we hope to be able to compare our models in the future.

I would like to change the focus of this issue, if you don't mind, to have a discussion on how we should proceed to indicate "how much" a pattern is similar to another one. Since SARI and CHIN mostly don't create new entities but design them in a specific way, it would be important to establish a protocol to identify the level of correspondence between different patterns. We would also like to do the same exercise with Linked.Art. In addition, it would be nice to create a concrete link between the patterns to ensure alignment when one of the patterns is modified.

This was something that CHIN has already in mind for other standards like Cataloguing Cultural Object (CCO) but I don't think it's the same kind of compliancy. In this case is to identify a field that can properly or not feed our model where the link with ZARI is more to determine if the semantics is aligned.

Of course, this will be easier when CHIN will release its model (#40)

[VladimirAlexiev]

A good practical acid test: can the same SPARQL query run against one and the other model without changes?

[Habennin]

Dear all,

This is a community wide concern of adopters of CIDOC CRM. As many larger scale, long term projects by keystone institutions adopt the CRM language to implement their semantic data strategies, the issue of ensuring practical implementation consistency (not just theoretical model consistency) becomes more and more urgent.

I agree with. @VladimirAlexiev that a good start is to see if the same sparql against different end points returns data and the same kind of data.

The publication of the SARI models in the format they are presently in was done to try to make a publicly accessible and hopefully easily digestible/readable presentation of how to carry out some common mapping tasks for some common entities and their properties.

Ideally, as a community we need to find/build the kind of tooling that would allow us to build consistent 'ontology profile' documentation/code that follows the same patterns to enable easy comparison of actual implementations.

One project going in this direction is the OntoMe platform: http://ontome.dataforhistory.org developed by LARHRA. At this point it is a tool for managing co-creation of ontology extensions, but aims also to be tool for defining ontology profiles.

Ultimately, it would be great to use something like ShaQL in order to be able to explicitly specify model patterns and validate model compliance.

[VladimirAlexiev]

You mean SHACL. And there is another shape language to consider: SHEX. I think it is slightly higher

[Habennin]

Indeed