Fix RAZU "is onderdeel van" issues
Opened this issue · 0 comments
Now: Showing "Gerechtsbestuur Amerongen, Elst en Ginkel" many times.
To preview: prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> prefix mdto: <http://www.nationaalarchief.nl/mdto#> select * where { ?s <http://www.nationaalarchief.nl/mdto#isOnderdeelVan> ?o. ?o mdto:begripLabel ?oLabel . } limit 1000 on https://data.razu.nl/razu/PoCAmerongen2024/sparql.
Wietse mentioned on June 19th that the SPARQL should instead look more like this:
prefix rdfs: http://www.w3.org/2000/01/rdf-schema# prefix mdto: http://www.nationaalarchief.nl/mdto# select * where { ?s http://www.nationaalarchief.nl/mdto#isOnderdeelVan/http://www.nationaalarchief.nl/mdto#identificatie/http://www.nationaalarchief.nl/mdto#identificatieKenmerk ?o. ?o mdto:naam ?oLabel . } limit 1000
Unfortunately we can not do any hops with elastic, it uses a completely "flat" index so only knows of "isOnderdeelVan" at the root level of the node and nothing more. We ran into similar issues with the The Utrecht Archives dataset earlier, but have no support (yet) for tweaking elastic indexing settings at Triply.
Even if I use a custom SPARQL query for rendering the "hierarchy" view for RAZU, the filter options would still be off as those are all handled by elastic.
Afraid we have to wait for more control over the elastic indexing process in order to resolve these issues. (either by self-hosting the triple store + elastic, or by Triply giving us more control over this)