Should we follow the principle of value for a property of Wikidata?

[candlecao]

The principle is generally:
In Wikidata, almost a property is actually either linking another URI(object property) or linking just literal value(data property).

For example, the value of property wdt:P136 genre are all URI, as you can refer by query:

PREFIX wd: <http://www.wikidata.org/entity/>
PREFIX wdt: <http://www.wikidata.org/prop/direct/>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

SELECT ?type (COUNT(?genre) AS ?count)
WHERE {
  ?item wdt:P136 ?genre .
  BIND(
    IF(
      isIRI(?genre),
      "URI",
      IF(
        isLiteral(?genre),
        "String",
        "Other"
      )
    ) AS ?type
  )
}
GROUP BY ?type

After executing this above, you see the count of URI is 1961941, but no even one literal data.

However, in our reconciliation task, we use string as the value of P136. Does that matter?

[fujinaga]

I don’t understand your question but have you looked into “music genre” Q188451?

[candlecao]

Thanks, Ich. I will look into it further.

[fujinaga]

Have you seen this https://www.wikidata.org/wiki/Wikidata:WikiProject_Music/Lists/Instance_of_music_genre?

[dchiller]

However, in our reconciliation task, we use string as the value of P136. Does that matter?

So are you saying that after reconciliation, the object of triples with predicate P136 are all literals? For something like genre, where we know that there are genre items in wikidata, it seems to me like we didn't actually reconcile those values...

[ahankinson]

It’s not really “Linked Data” if there’s no link… 😁

[candlecao]

However, in our reconciliation task, we use string as the value of P136. Does that matter?

So are you saying that after reconciliation, the object of triples with predicate P136 are all literals? For something like genre, where we know that there are genre items in wikidata, it seems to me like we didn't actually reconcile those values...

I think I partially conveyed my meaning to you. To clarify:

P136 is actually an "object property" in Wikidata.
But we use P136 as a "data property" in our reconciliation.
In our reconciled linked data, for genre, we use another property as object property instead, which is http://www.wikidata.org/prop/direct/P8052, called MusicBrainz genre ID.

Does that matter if we mix up object property and data property. I am not sure if this will interfere with the future project.

[dchiller]

In my mind, there should be a Wikidata URI as the object of P136 in our reconciled data. If there isn't, my contention is that we haven't reconciled the data against Wikidata.

Am I understanding?

[candlecao]

Yes, I think you basically understood my meaning. But there is some trivial stuff to be explained. Thanks.

[fujinaga]

Yes, there will be a Wikidata URI as the object for P136 unless the original genre URI is the Musicbrainz genre ID, in which case, we will use P8052 instead of P136.

[Yueqiao12Zhang]

Using this convention, since in my schema, I link the graphs using MBIDS but not Wikidata IDs, we must to through

  graph TD;
      A[Rumors release-group]--P8052--->C[797e2e85-5ffd-495c-a757-8b4079052f0e];
      C--P136-->D[https://www.wikidata.org/wiki/Q484641];

Loading

Since we have to track using MBID before we can access the QID for that object.

[dchiller]

Ok, but I don't think that's what P8052 means. It looks to me like the subject of P8052 is a genre and the object is a literal MusicBrainz ID.

[dchiller]

PS. I edited your comment so the mermaid graph would display.

[Yueqiao12Zhang]

Yes, that's where we were having problems because we were deciding to use P136 or P8052. Now we think we should remove the genre_name column because we don't need it. The problem is which property we should use when linking a subject to genre MBID.

[dchiller]

The problem is which property we should use when linking a subject to genre MBID.

A release group, for example, will have a genre. That genre may/will have multiple reconciled identifiers (it'll have a Wikidata ID, it'll have a MusicBrainz ID, maybe one day it'll have a "Dylan's great big ontology" ID) ... but I don't think there's any question of which of P136 or P8052 to use to link a release group (or a recording, or a song, etc) with a genre! It's P136!

[Yueqiao12Zhang]

So that's what we decided to do:

  graph TD;
      A[https://musicbrainz.org/release-group/416bb5e5-c7d1-3977-8fd7-7c9daf6c2be6]--P136--->C[https://musicbrainz.org/genre/797e2e85-5ffd-495c-a757-8b4079052f0e];
      C--P2888-->D[https://www.wikidata.org/wiki/Q484641];

Loading

Then we use the reasoner in Virtuoso to complete the relation:

  graph TD;
      A[https://musicbrainz.org/release-group/416bb5e5-c7d1-3977-8fd7-7c9daf6c2be6]--P136--->D[https://www.wikidata.org/wiki/Q484641];

Loading

[dchiller]

FYI, use "```mermaid" to get the graph to display.

[dchiller]

So what is in our graph? The first, the second, or both?

[Yueqiao12Zhang]

My code makes the first, and in Virtuoso, Junjun converts it to the second.

[fujinaga]

The reason I mentioned P8052 is because I wanted to ensure that there's an indication that the genre was assigned by MusicBrainz. I'm not sure what's the best way to do this. This is to differentiate from some other music database that uses MusicBrainz genre ID for their assigned genre.

[candlecao]

OK, got it. We will handle this issue more clearly.
We may talk about it face to face.