DBpedia Fr structure

Metadata

DBpedia Fr provides a complete description of itself using the void, dcat, sd and prov ontologies.

That description is composed of two layers :

  • The first one is static : you can access it directly through this page or via http://fr.dbpedia.org/void/ or http://fr.dbpedia.org/.well-known/void. It contains all the information concerning the maintenance and the creation of the data: its contributor, the data source, the code that produced this one... They are loaded in a first named graph: <http://fr.dbpedia.org/graph/metadata>. Moreover, it informs about how the DBpedia Fr data are structured. ndeed, we have chosen to partition the triple into different named graphs according to the source and process they result.
  • The second is dynamic: they are computed at each loading or during an update. It notifies, for example, the date of theses. But above all, it contains statistical information about the fact of each graph.

Named graphs

DBpedia Fr contains many triples, they spread out named graphs. Each graph is the result of a different extraction process, as available on our Databus collection.

By convention their URI is defined as follows: <http://fr.dbpedia.org/graph/$nom_artefact> You can specify easily which name graph you want to query with a FROM statement. For example, if you want to get the type of a particular resource, it is better to define that you target the dedicated named graph, rather than letting Virtuoso lookup for a specific property in the entire knowledge base.: 

                                    
                SELECT ?type FROM <http://fr.dbpedia.org/graph/dbpedia_mappings_instance-types>
                WHERE 
                {
                  dbpedia-fr:Paris a ?type

                }

The performance gain in time is not significant on that simple query, but when you work on a complex query it will make sense.

You can visualize below these named graphs depending on the number of triples they contribute to DBpedia Fr. Mouse over the bubbles for reading some information about each graph. Click on it if you want to fix the data in the right-hand table.

metadawikidatagenerictextmappingsstatisticsother

Properties Value
Name
Nb triples
Nb properties
Nb entities
Nb classes
Description
Generated at
Process
DataDump

Browsing the metadata

Metadata are also available into the DBpedia endpoint in the named graph : <http://fr.dbpedia.org/graph/metadata>. therefore it make possible to querying them via SPARQL.

Description du endpoint :

              
              SELECT  ?p  ?o  FROM <http://fr.dbpedia.org/graph/metadata>
              WHERE 
              {
               <http://fr.dbpedia.org/> ?p  ?o.  
                FILTER (!isBlank(?o))
              }

Note that we choose to not diplaying the blank nodes (Filter !isBlank)

Named graph list

For getting all the named graph, it's possible to use the following query :

                    
            SELECT ?o FROM <http://fr.dbpedia.org/graph/metadata>
            WHERE 
              {
               <http://fr.dbpedia.org/> sd:namedGraph  ?o
              }

Getting metadata specific to a named graph

If you want to learn more about a given named graph, for example this one : <http://fr.dbpedia.org/graph/dbpedia_generic_infobox-properties>, you just have to use this query :

                            
            SELECT ?p  ?o  FROM <http://fr.dbpedia.org/graph/metadata>  
            WHERE {
                <http://fr.dbpedia.org/graph/dbpedia_generic_infobox-properties>  ?p  ?o.
                FILTER (!isBlank(?o))
            }

You can access via this query to all the following propertie

  • <http://www.w3.org/ns/prov#wasGeneratedAtTime> and <http://schema.org/datePublished> : notify the update date
  • <http://rdfs.org/ns/void#dataDump> : related to the dump that we are hosting
  • <http://www.w3.org/ns/prov#hadPrimarySource> :refer to the databus artefact used
  • <http://rdfs.org/ns/void#vocabulary> : list the voculary used in the named graph
But you can also consult global statistics :
  • <http://rdfs.org/ns/void#classes> : number of object by classes in the named graph
  • <http://rdfs.org/ns/void#triples> : number of triple in the named graph
  • <http://rdfs.org/ns/void#entities> : number of different entities in the named graph
  • <http://rdfs.org/ns/void#properties> : number of properties used in the named graph

Access to fine grained statistics

In addition to global statistics, it is possible to have access to fine statistics computed on each named graph. Most of them aren't reachable using a COUNT statement, because they require too much computation time. We pre-calculated and made them accessible using the properties partitions and class partitions, proposed by the void vocabulary. We recommend, therefore, that you do not re-calculate these statistics yourself using a COUNT clause.

Properties partitions

Give you access for each property of a named graph to :

  • the number of triple using it/li>
  • the number of distinct object related to them
  • the number of distinct subject related to them

                    
    SELECT  ?p ?nb_triples ?nb_subj ?nb_obj FROM <http://fr.dbpedia.org/graph/statistics> 
    WHERE {
        
        {
             <http://fr.dbpedia.org/graph/dbpedia_generic_infobox-properties>  void:propertyPartition 
             [ void:triples ?nb_triples ; void:property ?p ;].
             <http://fr.dbpedia.org/graph/dbpedia_generic_infobox-properties>  void:propertyPartition 
             [ void:distinctSubjects  ?nb_subj ; void:property ?p ;].
             <http://fr.dbpedia.org/graph/dbpedia_generic_infobox-properties>  void:propertyPartition 
             [ void:distinctObjects  ?nb_obj ; void:property ?p ;].
        }        
           
    }

Class partitions

Give you access for each class of a named graph to :

  • the number of triples by type
  • the number of objects related to this type
  • the number of subjects related to this type
  • the number of properties linking at least one object of this type 

                    
  SELECT  ?g ?c ?nb_triples ?nb_subj ?nb_obj ?nb_classes ?nb_prop FROM <http://fr.dbpedia.org/graph/statistics> 
  WHERE {
      
   {
        <http://fr.dbpedia.org/graph/dbpedia_generic_infobox-properties>  ?g  void:classPartition
         [ void:triples ?nb_triples ; void:class ?c ;].
        <http://fr.dbpedia.org/graph/dbpedia_generic_infobox-properties>  ?g void:classPartition 
         [ void:distinctSubjects  ?nb_subj ;void:class ?c;].
       <http://fr.dbpedia.org/graph/dbpedia_generic_infobox-properties>  ?g  void:classPartition 
         [ void:distinctObjects  ?nb_obj ; void:class ?c ;].
       <http://fr.dbpedia.org/graph/dbpedia_generic_infobox-properties>  ?g  void:classPartition 
         [ void:classes  ?nb_classes ; void:class ?c ;].
        <http://fr.dbpedia.org/graph/dbpedia_generic_infobox-properties>  ?g  void:classPartition 
         [ void:properties  ?nb_prop; void:class ?c ;].

    }           
         
  }