Missing `rdf:type`s in HTML view and in some SPARQL answerse

[jmkeil]

Issue validity

Could not check, due to DIEF ERROR 502 response.

Error Description

Many resources are missing some rdf:type statements.

Pinpointing the source of the error

It affects the SPARQL (http://dbpedia.org/sparql) endpoint and the HTML (http://dbpedia.org/resource/…) representation. I did not check further sources.

Details

It seems to affect many different resources. I will point it out on the example of resources representing astronauts.

• Astronauts should have the rdf:type dbo:Astronaut, but some do not, e.g. https://dbpedia.org/page/Anne_McClain

• some other resources are not affected, e.g. https://dbpedia.org/page/John_Young_(astronaut)

• Wikipedia sources of affected and not affected resources contain the Infobox astronaut

  • affected: https://en.wikipedia.org/wiki/Anne_McClain
  • not affected: https://en.wikipedia.org/wiki/John_Young_(astronaut)

• the Infobox astronaut is properly mapped: https://mappings.dbpedia.org/index.php/Mapping_en:Infobox_astronaut

• dbo:Astronaut defined on https://mappings.dbpedia.org/index.php/OntologyClass:Astronaut as equivalent class of http://www.wikidata.org/entity/Q11631

• affected and not affected resources have the rdf:type http://www.wikidata.org/entity/Q11631

• http://dbpedia.org/sparql gives inconsistent answers about the types:

  SELECT * WHERE {
  	<http://dbpedia.org/resource/Anne_McClain> rdf:type ?type
  }
  

  -> dbo:Astronaut not contained, wikidata:Q11631 contained

  DESCRIBE <http://dbpedia.org/resource/Anne_McClain>
  

  -> dbo:Astronaut not contained, wikidata:Q11631 contained

  ASK {
  	<http://dbpedia.org/resource/Anne_McClain> rdf:type dbo:Astronaut .
  }
  

  -> true

  ASK {
  	<http://dbpedia.org/resource/Anne_McClain> rdf:type <http://www.wikidata.org/entity/Q11631> .
  }
  

  -> true

  SELECT ?case (COUNT(DISTINCT ?entity) AS ?COUNT) WHERE {
  	?entity a <http://www.wikidata.org/entity/Q11631> .
  	FILTER NOT EXISTS { ?entity a dbo:Astronaut . }
  	BIND("only WD" AS ?case)
  }
  GROUP BY ?case
  

  -> 0

  SELECT ?case (COUNT(DISTINCT ?entity) AS ?COUNT) WHERE {
  	?entity a dbo:Astronaut .
  	?entity a <http://www.wikidata.org/entity/Q11631> .
  	BIND("DBO and WD" AS ?case)
  }
  GROUP BY ?case
  

  -> 713

  SELECT ?case (COUNT(DISTINCT ?entity) AS ?COUNT) WHERE {
  	?entity a dbo:Astronaut .
  	FILTER NOT EXISTS { ?entity a <http://www.wikidata.org/entity/Q11631> . }
  	BIND("only DBO" AS ?case)
  }
  GROUP BY ?case
  

  -> 26

  SELECT ?case (COUNT(DISTINCT ?entity) AS ?COUNT) WHERE {
  	?entity a dbo:Astronaut .
  	BIND("DBO" AS ?case)
  }
  GROUP BY ?case
  

  -> 739

  SELECT ?case (COUNT(DISTINCT ?entity) AS ?COUNT) WHERE {
  	?entity a <http://www.wikidata.org/entity/Q11631> .
  	BIND("WD" AS ?case)
  }
  GROUP BY ?case
  

  -> 713

As the HTML rendering is consistent with the DESCRIBE and the not aggregating SELECT queries, this likely is an SPARQL endpoint error. However, an extraction can also not be excluded completely.

[RitikaSen27]

Hello,

I’m interested in looking into this issue. From the examples and queries provided, it seems there’s an inconsistency between rdf:type materialization for dbo:Astronaut and the equivalent Wikidata class (wd:Q11631), where ASK queries succeed but DESCRIBE and SELECT results don’t consistently expose the DBO type.

I’d like to investigate whether this is due to SPARQL endpoint inference/materialization behavior, ontology equivalence handling, or an extraction-side issue, and try to narrow it down with smaller test cases.

Please let me know if it’s okay to proceed with a focused investigation and report findings (and a fix, if applicable).I would definitely want to work on this issue.

Thank you.

[PritamP20]

Hey @jmkeil,

I went through the issue and the details you shared — I’d like to work on this one. From what I can see, the inconsistency where ASK works but SELECT and DESCRIBE don’t return dbo:Astronaut might be due to how the SPARQL endpoint handles inference or materialization between dbo:Astronaut and the equivalent Wikidata class.

I’ll start by reproducing this on a smaller dataset and try to pinpoint whether it’s coming from the endpoint behavior or extraction side. Could you please assign this issue to me so I can dig into it?

[rakshaak29]

Hi @jmkeil ,
I went through this issue and tried to understand it by looking at the DBpedia HTML view for a few astronaut pages.
I compared Anne McClain and John Young (astronaut). Both use an astronaut infobox, and the astronaut mapping exists. However, Anne McClain is missing rdf:type dbo:Astronaut, while John Young correctly has rdf:type dbo:Astronaut in the DBpedia HTML view.
This suggests that the mappings and ontology are working as expected, but the extraction pipeline is behaving inconsistently. It appears that the ontology type generated via infobox mappings is either not created in some cases or is being removed during post-processing, possibly when Wikidata types are present.
I am currently exploring the extraction framework to investigate where this behavior occurs. Please assign this issue to me so I can continue working on it and propose a fix.
Thanks!