Completion of the knowledge graph with DBpedia and Wikidata about music_kg

[adrigrillo]

With regarding the connection of the existent artist to the new genres, this task is difficult to generalize because of the high probability of mixing information of two different artists that have the same name. An example of a query that show the problem using DBpedia is:

PREFIX mo: <http://purl.org/ontology/mo/>
PREFIX dc: <http://purl.org/dc/elements/1.1/>
PREFIX dbo: <http://dbpedia.org/ontology/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> 
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

select ?db_artist_name ?artist ?db_genre_name ?db_artist
where {
    service <http://dbpedia.org/sparql> {
        ?db_artist foaf:name ?db_artist_name_en .
        bind(str(?db_artist_name_en) as ?db_artist_name)
        ?db_artist dbo:genre ?db_genre .
        ?db_genre rdfs:label ?db_genre_name_en .
        filter(langMatches(lang(?db_genre_name_en), "en")).
        bind(replace(lcase(str(?db_genre_name_en)),"[\\(.*\\)]","") AS ?db_genre_name)
    }
    filter exists {
        ?genre a mo:Genre .
   		?genre rdfs:label ?db_genre_name .
    }
    ?artist rdfs:label ?db_artist_name
}

And its result the following table, where it can be seen that more than a artist is matching because of the name.

?db_artist_name	?artist	?db_genre_name	?db_artist
Asia	http://musicbrainz.org/artist/8f695099-c714-4ad0-8888-36eca37fee84	art rock	http://dbpedia.org/resource/Asia_(band)
Asia	http://musicbrainz.org/artist/56b087c0-d336-4e4d-a0fe-67a04708bb13	art rock	http://dbpedia.org/resource/Asia_(band)
Asia	http://musicbrainz.org/artist/2841cb74-4ad3-4577-828c-8348b4c814f9	art rock	http://dbpedia.org/resource/Asia_(band)
Asia	http://musicbrainz.org/artist/9e9621f7-88f1-4f78-9b9c-3548b6999df8	art rock	http://dbpedia.org/resource/Asia_(band)

A possible solution is to use all the possible information of the existent group to afterwards query the external source.

[adrigrillo]

Wikidata

The search of new genres in Wikidata returns 1217 new genres that are not contained already contained in the graph. This includes genres that have the same name and the ones that already have been previously linked to Wikidata (#7).

The query that generates the triples that will be added to the graph can be seen in this file.

DBpedia

The search in DBpedia returns 389 new genres that are not in already in the graph, counting with the ones added from Wikidata. The linking done in the previous step (#7) is took into account also. Moreover, in DBpedia some strings of the genres contains text in parenthesis, generally specifying that is a music genre, as flamenco rock (genre). This text between the parenthesis have been removed.

The query that generates the triples that will be added to the graph can be seen in this file.