Missing DBpedia Ontology types in DBpedia live SPARQL endpoint

Services and Applications DBpedia Live
1

Hi,

We’ve noticed that the types for an entity does not include the Dbpedia ontology when queried via the ‘live.dbpedia.org’ Sparql endpoint, only if we query ‘dbpedia.org’ ? Why is this?

Regards
Chris

2

Hi @chris.rafferty

DBpedia.org uses a static ontology which is dumped together with the dataset.

Live.dbpedia.org gets the ontology as part of the live extraction scheme which seems to have some differences in generating the ontology triples.

I will check with the Live development team to see what could be wrong here and report back to you.

Patrick

3

Hi Patrick,

Did you get any response from the Live development team?

Regards
Chris

4

@chris.rafferty I am unsure how to test this. The types are different per se. The main endpoint has data from 2017 and will be updated in snapshots. Live is continuous and reflects the mappings and the changes in Wikipedia.

Could you be more concrete, e.g. give a SPARQL query, examples or statistics? Because I can not estimate whether it is an error or system inherent.

5

Hi,

Sure, however there is a restriction on the number of links, myself as a new user, can post. So here’s the query with the resource link missing. I’ll follow-up with a couple of examples to substitute in, the query does not produce results on live.dbpedia.org/sparql

PREFIX ont: http://dbpedia.org/ontology/
SELECT ?type
WHERE {
PUT RESOURCE LINK HERE a+ ?type
filter (strstarts(str(?type),str(ont:)))
}

Regards

Chris

6

Here’s the examples:

http://dbpedia.org/resource/London
http://dbpedia.org/resource/HMS_Hood

7

Thanks for removing the restrictions here’s a populated query:

PREFIX ont:<http://dbpedia.org/ontology/>
SELECT ?type
WHERE {
<http://dbpedia.org/resource/London> a+ ?type
filter (strstarts(str(?type),str(ont:)))
}

Also we have noticed one resource on live that does have the dbpedia ontology types:

<http://dbpedia.org/resource/Douglas_Adams>

8

my educated guess is that the state of the resource got corrupted. the live sync mechanism is very tricky. a lot can go wrong. you can e.g. see at the moment multiple revisionids. that is an inconsistent state which should never occur in a perfect world. there is a failover procedure for things like this. every 90 days or after 10 deltas were applied ( @kurzum correct my if i am wrong and the config is now different) the resource is deleted completely and inserted again completely from scratch (no deltas are applied)

ont:wikiPageRevisionLink
http://en.wikipedia.org/w/index.php?title=London&oldid=937793183
95	
ont:wikiPageRevisionLink
http://en.wikipedia.org/w/index.php?title=London&oldid=937793725
9

however for http://dbpedia.org/page/HMS_Hood there is only one revision id but also a dbo:Work type (at the moment). @kurzum debuggin can be done in several ways. start DIEF in server mode on live server and look what the extraction returns for London and HMS given the revision ids.
moreover grep the delta files for the resources and then grep again for types to see whether they occur at all in input.