search

FAIRsFAIR / FAIRSemantics

MIT License
7 stars 1 forks source link

P-Rec. 12: Semantic mappings between the different elements of semantic artefacts should be published in machine-readable formats. (D2.5) #39

Open GCoen1 opened 3 years ago

GCoen1 commented 3 years ago

As we discussed in P-Rec. 10, semantic artefacts are often developed to describe a specific aspect of a scientific domain. Despite this reduced scope, several models of the same aspects can co-exist. They are either developed de novo or developed as a part of another ontology.

This leads to potential for divergence, often due to a lack of knowledge regarding existing semantic artefacts. To manage the impacts of such divergence, it is necessary to allow mapping of relations between the elements of such semantic artefacts (ontological alignments).

In many cases, these mappings can be based on existing relations (such as sameAs from SKOS). However, mappings can become complex especially when considering logical relations. Risks include representing content drift, as well as context insensitive use of semantic artefacts, and there are no common descriptions for such complex mappings. Mappings are often created by individuals for satisfying a specific need. Information regarding the provenance and usage of these mappings are of importance for any practitioner who would be interested in reusing them.

Principle I3 states that “I3. (Meta)data include qualified references to other (meta)data” - the “qualified references” referred to here are essentially mappings. This recommendation aims at highlighting a gap in relation to the achievement of principle I3, and to emphasize the need for harmonisation. It will require the implementation of machine readable descriptions of mappings in order to foster interoperability.

dr-shorthair commented 2 years ago

sameAs is from OWL, not SKOS. SKOS has exactMatch for mappings.

matentzn commented 2 years ago

As I mentioned elsewhere, you may be interested in SSSOM as a standard way to publish fair mappings. It comes with simple table-based format, and a few standard serialisations like RDF and JSON-LD. In the next 10 days or so, we will also upload a pre-print on arxiv.

matentzn commented 2 years ago

Sorry, its been a while, my comment is redundant with https://github.com/FAIRsFAIR/FAIRSemantics/issues/40