Skip to content

DBPedia

Jump to bottom
Tim L edited this page May 8, 2014 · 50 revisions

How to map Wikipedia to DBPedia

For example, http://en.wikipedia.org/wiki/GOES_7 uses the "spaceflight" infobox. You can figure out what infobox it uses by clicking the page's "edit" tab to see the template syntax {{Infobox spaceflight in the source code. That infobox is defined at Template:Infobox_spaceflight and Wikipedians use it to describe satellites. When viewing the infobox at http://en.wikipedia.org/wiki/GOES_7, you can see the COSPAR ID is 1987-022A. We'll track that property-value through the DBPedia's mapping.

DBPedia's corresponding URI for http://en.wikipedia.org/wiki/GOES_7 is http://dbpedia.org/resource/GOES_7 (this greasemonkey script can help jump the gap from Wikipedia to DBPedia). Resolving DBPedia's URI shows dbpprop:cosparId 1987 (xsd:integer). The property's URI is http://dbpedia.org/property/cosparId, which doesn't really resolve to anything useful. DBPedia's mapping is controlled by editing http://mappings.dbpedia.org/index.php/Mapping_en:Infobox_spaceflight. To control the mapping, edit the page to put an occurrence of the http://mappings.dbpedia.org/index.php/Template:TemplateMapping.

TODO: http://mappings.dbpedia.org/index.php/OntologyProperty:CosparId. Keep an eye on "What links here" from a property page, so you know which other infobox mappings map to the DBPedia property.

Classes can be added to the DBPedia ontology by putting a Class template into e.g. http://mappings.dbpedia.org/index.php/OntologyClass:ArtificialSatellite (for http://dbpedia.org/ontology/ArtificialSatellite).

DBPedia's Mapping documentation

dbp vs. dbo

20 March 2014

For more details on how DBpedia Live works I suggest you read the related publication [1]. Regarding the updates, we offer the dbpintegrator tool [2] that can sync your local triple store with DBpedia Live. This is the suggested option if you plan to make heavy use of the endpoint.

Once you have the syncing done,we now provide the following properties:

  • dbpedia-owl:wikiPageExtracted
  • dbpedia-owl:wikiPageModified that you can use to check for newly extracted pages.

DBpedia Live gets all information from the articles except from abstracts and images. Abstracts will be supported soon while for images we cannot offer real-time updates due to the current image Extractor architecture.

Best, Dimitris

[1] Mohamed Morsey, Jens Lehmann, Sören Auer, Claus Stadler, Sebastian Hellmann, (2012) «DBpedia and the live extraction of structured data from Wikipedia", Program: electronic library and information systems, Vol. 46 Iss: 2, pp.157 – 181 http://jens-lehmann.org/files/2012/program_el_dbpedia_live.pdf [2] http://sourceforge.net/projects/dbpintegrator/

23 Jan 2014

you can improve / increate the data that are extracted from Wikipedia by better mapping the infobox to the DBpedia ontology [1] [2] This is a crowdsourced process that needs regular updates due to changes in Wikipedia templates

Once you improve the mappings, DBpedia Live [3] will reflect all the changes (within a short time) but dbpedia.org will be updated on the next static release.

Best, Dimitris

[1] http://mappings.dbpedia.org/index.php/Mapping_en:Infobox_German_location [2] http://mappings.dbpedia.org/server/templatestatistics/en/?template=Infobox_German_location [3] http://live.dbpedia.org

8 Dec 2013

  • dbp: is a legacy namespace for properties. (i.e. "raw infobox data")
  • dbo: is the current namespace for properties.

So, universities in countries

prefix dbpedia: <http://dbpedia.org/resource/>
prefix dbo:     <http://dbpedia.org/ontology/>
prefix dbp:     <http://dbpedia.org/property/>

select ?country count(distinct ?university) as ?count
where {
           ?university dbo:type dbpedia:Public_university ;
            dbp:country ?country .
           filter (isIRI(?country))
}
group by ?country
order by desc(?count)

7 Dec 2013

If you restrict yourself to the DBpedia ontology, you'll get much more sensible results. E.g., if you execute this query on the DBpedia SPARQL endpoint (http://dbpedia.org/sparql) (which includes the prefix dbpedia-owl: http://dbpedia.org/ontology/):

 select ?country (count(?university) as ?count)
 where {
   ?university a dbpedia-owl:University
   optional{ ?university dbpedia-owl:country ?country}
 }
 group by ?country

you get much better results. For more about the differences, see this StackOverflow question (http://stackoverflow.com/q/17237144/1281433), and some of the DBpedia documentation that the answer links to.

Two other cases where the noise in the raw infobox data affected people's results:

http://stackoverflow.com/q/16977418/1281433 http://stackoverflow.com/q/18834426/1281433

-Josh Taylor

DBPedia tables

12 Dec 2013

Petar Ristoski set up a web page where all classes from the DBpedia ontology are available for download as separate .csv and .json files.

14 Nov 2013

Announced the first version of DBpedia as Tables

DBPedia Live

12 Dec 2013

http://live.dbpedia.org announced update

DBPedia dumps

older

http://dbpedia.org/downloads states the dumps we use for the latest version and http://downloads.dbpedia.org/ holds a dump archive

for the latest updates you can refer to live.dbpedia.org (might be unresponsive due to maintenance)

-Dimitris

Lookup service

http://lookup.dbpedia.org/

the service is working as expected. The results are ordered by RefCount (i.e., how often the respective Wikipedia article is linked from another one) in descending order. By default, the service returns the top 5 results, and http://dbpedia.org/resource/Building is simply not among those.

Try http://lookup.dbpedia.org/api/search.asmx/KeywordSearch?QueryString=building&MaxHits=100, and you will find what you are looking for.

Clone this wiki locally