Linksammlung Statistische Maschinelle 

UIMA wrapper for the German Chunker

Here you can download a UIMA wrapper for the German Chunker by Helmut Schmid and Sabine Schulte im Walde.
The chunker annotates recursive noun chunks and also outputs case information for each chunk. It is available at It runs under Linux and Solaris.

The input types needed by the chunker are Token and Sentence, as produced by the Tokenizer and SentenceSplitter annotators from the Darmstadt Knowledge Processing Repository (DKPro).
The chunker outputs the type Chunk, which has three features:

  • pos: category of the chunk
  • case: the chunk's case (nominative, dative, genitive, accusative)
  • prepHead: the prepositional head of the chunk, if any.

Download: unipotsdam-german-chunker-ae.tar.gz


  • Download the above package and unpack it. This will create the new directory unipotsdam-german-chunker-ae/
  • Download the binary package of the chunker from IMS Stuttgart and extract it to unipotsdam-german-chunker-ae/resources/GermanChunker/. Also copy the executable binary lopar (from the LoPar package available at the LoPar homepage at IMS Stuttgart) into this directory.
  • Change to the directory unipotsdam-german-chunker-ae/desc/ and edit the XML descriptor file GermanChunkerAnnotator.xml. You have to configure the paths in lines 22, 23, and 71 according to your settings.
  • Before running a CPE with the chunker, don't forget to add the directory unipotsdam-german-chunker-ae/bin/ to the UIMA classpath.

© 1999-2010 Peter Kolb   |  5.2.2010