Missing mappings for exact_match and close_match (skos)

[BernhardKoschicek]

In the linked.art context file I cannot find exact_match or close_match mapped, even though they are listed in the name collision section of the Linked Art API documentation: https://linked.art/api/1.0/json-ld/.

I have a local copy of an older linked.art JSON-LD context file (also version 1.1) where these terms are mapped:

{
  "exact_match": {
    "@id": "skos:exactMatch",
    "@type": "@id",
    "@container": "@set"
  },
  "close_match": {
    "@id": "skos:closeMatch",
    "@type": "@id",
    "@container": "@set"
  }
}

It appears that these mappings were removed in commit c977ddd.

My question is: Should I be using skos:exactMatch and skos:closeMatch directly instead of exact_match and close_match within the Linked Art context? Or was the removal of these terms from the Linked Art context file an oversight?

I have to say, I'm quite new to the LOD universe, and I'm just trying to map CIDOC CRM data with your context file to RDF.

[azaroth42]

Well spotted, thank you!

No, the removal from the context was intentional, but the oversight was /not/ removing them from the table in the json-ld page.

Instead of skos:exactMatch we minted our own term la:equivalent to avoid the inference issue with CRM, being that CRM equates skos:Concept with crm:E55_Type, so if you said that <person1> skos:exactMatch <person2> both people would also become E55 Types. The issue for that discussion is #307

Also schema:genre should be removed. I'll double check the others as well.

[BernhardKoschicek]

Thank you, then I have to ask, do you also have an equivalent for skos:closeMatch?

[azaroth42]

There isn't. You can add new mappings in a new context, but ... do you really need both exact and close match? Our experience is that there's no functional value in something that's basically saying: "This might be the same, it's kinda close, but not quite..."

[BernhardKoschicek]

We, in OpenAtlas, use close and exact match this way:

• Close match: Concepts (here a dataset in OpenAtlas and a dataset in an external reference system) are sufficiently similar, therefore they can be used interchangeably in some information retrieval applications
• Exact match: High degree of confidence that the concepts (a dataset in OpenAtlas and a dataset in an external reference system) can be used interchangeably
  https://manual.openatlas.eu/entity/reference_system.html#precision

So for example the type "Roman coin" is connected to "coin" in Wikidata and "coin" in getty with a close match, because it is a coin, but a special kind of coin which is not represented in wikidata or getty.

But of course, the outcome of #307 is coherent, and maybe we should think about that in our model, since we also use it for persons and places the same way as for types.

[aisaac]

@BernhardKoschicek Wouldn't the relation between "Roman coin" and the Wikidata and AAT coins be skos:broadMatch instead of skos:closeMatch? In this case you could use skos:broader, which is in the Linked Art context (skos:broadMatch being a sub-property of skos:broader, it's fine to use the latter in stead of the former)

[azaroth42]

Call on 2025-04-30: Doesn't make sense to fix the context -- would be a regression -- the documentation is wrong.
Have an errata as patch without removing, and then a minor version to remove them. Explain la:equivalent as the approach we use.