The AKSW group is pleased to announce the official release of DBpedia Live. The main objective of DBpedia is to extract structured information from Wikipedia, convert it into RDF, and make it freely available on the Web. In a nutshell, DBpedia is the Semantic Web mirror of Wikipedia.
Wikipedia users constantly revise Wikipedia articles with updates happening almost each second. Hence, data stored in the official DBpedia endpoint can quickly become outdated, and Wikipedia articles need to be re-extracted. DBpedia Live enables such a continuous synchronization between DBpedia and Wikipedia. There was a preliminary PHP based DBpedia Live extraction framework available, which was, however, based on an older version of the DBpedia extraction framework and did not contain all necessary features to synchronise properly with Wikipedia.
The DBpedia Live framework has the following new features:
- Migration to the new Java/Scala DBpedia framework.
- Support of clean abstract extraction.
- Automatic reprocessing of all pages affected by a schema mapping change at http://mappings.dbpedia.org.
- Automatic reprocessing of pages that are not changed for more than one month. The main objective of that feature is to that any change in the DBpedia framework, e.g. addition/change of an extractor, will eventually affect all pages. It also serves as fallback for technical problems in Wikipedia or the update stream.
- Publication of all changesets.
- Provision of a tool to enable other DBpedia mirrors to be in synchronization with our DBpedia Live endpoint. The tool continuously downloads changesets and performs changes in a specified triple store accordingly.
- SPARQL-endpoint: http://live.dbpedia.org/sparql
- DBpedia-Live Statistics: http://live.dbpedia.org/livestats
- Changesets: http://live.dbpedia.org/liveupdates
- Sourcecode: http://dbpedia.hg.sourceforge.net/hgweb/dbpedia/extraction_framework
- Synchronization Tool: http://sourceforge.net/projects/dbpintegrator/files/
For further questions, please use the mailing-list at firstname.lastname@example.org. We thank our partners at the FU Berlin and OpenLink for their support.