关于统计DBpedia资源总量及SPARQL查询有效性的技术咨询

Hey there! Great question—totally get the confusion when you're starting out with SPARQL and DBpedia. Let's break this down clearly:

Your initial query doesn't equal total DBpedia resources #

The query you shared:

SELECT (count(distinct ?sub ) as ?amount) WHERE { ?sub ?predic ?obj . }

counts all unique URIs that appear as a subject in DBpedia triples. That includes way more than just the http://dbpedia.org/resource/ entities you’re targeting. For example, it will count:

• Properties from the http://dbpedia.org/property/ namespace
• Ontology classes from http://dbpedia.org/ontology/
• External URIs from linked datasets
• Even internal system-related URIs

So this number is significantly larger than the actual total of DBpedia’s core resources.

The correct way to count DBpedia resources #

To get only the resources under the resource/ namespace, use the STRSTARTS function to filter subjects by their URI prefix. Here’s a precise query:

SELECT (COUNT(DISTINCT ?sub) AS ?totalDBpediaResources)
WHERE {
  ?sub ?pred ?obj .
  # Filter to only include resources in the DBpedia resource namespace
  FILTER (STRSTARTS(STR(?sub), "http://dbpedia.org/resource/"))
  # Optional: Exclude category pages if you don't want them counted
  FILTER (!STRSTARTS(STR(?sub), "http://dbpedia.org/resource/Category:"))
}

This works because:

• The first filter ensures we only look at URIs starting with the DBpedia resource prefix
• The optional second filter removes category pages (these live in the resource namespace but are metadata, not core entities like people or places)

A quick precision tip #

If you want an even stricter count of "entities" (things like people, cities, or organizations), you could add a filter for entities with an ontology type (e.g., ?sub rdf:type dbo:Thing), but the namespace filter is usually sufficient for a general "total resources" count.

内容的提问来源于stack exchange，提问作者Xyz1337