DBpedia Mapping Server

DBpedia Mapping Server


The DBpedia Mapping Server provides statistics and other analytical information about the DBpedia ontology and mappings that are maintained in the DBpedia Mappings Wiki. (Note that this is distinct from the Virtuoso-based server which handles the DBpedia SPARQL and other Linked-Data-related access points.)


All the instructions below work on a Linux box. With any other operating system, you're pretty much on your own.

Preparation

 

  1. Check out the dump branch of the extraction_framework module from the DBpedia Mercurial repository. (Note that the branch may change in the future.)
  2. To speed up compilation, comment out or remove the line <module>scripts</module> in the parent pom.xml. If you don't need to generate the statistics files (see below), you can also remove the line <module>dump</module>.


To generate the mapping statistics, the server needs usage counts of the Wikipedia templates and their properties. Either copy these over from an existing server installation, or extract them:

 

  1. Download the latest Wikipedia dumps for all languages that have mappings in the mappings wiki.
  2. Run the DBpedia extraction framework with the configuration given in extraction.stats.properties:
    1. cd to the dump directory
    2. Execute ../clean-install-run extraction extraction.stats.properties &>extraction.2012-07-28.01.log & (adjusting the log file date and number, of course)
    3. TODO: anything else?
  3. Convert the extracted RDF triples to the statistics file format used by the server:
    1. cd to the server directory
    2. Execute ../clean-install-run stats &>stats.2012-07-28.01.log & (adjusting the log file date and number, of course)
    3. TODO: anything else?

Starting the server


To start the server:

 

  1. cd to the server directory
  2. change three or four lines in pom.xml:
    1. Enter the URL of the mappings wiki in <arg>http://mappings.dbpedia.org/</arg>.
    2. Use a proper password in the line <arg>CHANGE-THIS-PASSWORD</arg>. This password will be needed to put templates and template properties on the ignore list used by the statistics.
    3. Change <arg>../ontology.xml</arg> to a nonexistent file, e.g. <arg>do not use ../ontology.xml</arg> to force the server to download the current ontology from the mappings wiki.
    4. Change <arg>../mappings</arg> to a nonexistent folder, e.g. <arg>do not use ../mappings</arg> to force the server to download the current mappings from the mappings wiki.
  3. Run ../clean-install-run server &>server.2012-07-28.01.log & (adjusting the log file date and number, of course)

Stopping the server


Just kill the process. Usually, the log file will then show a Maven shutdown error message. Ignore it.