1. 23 Jul, 2018 1 commit
  2. 11 Aug, 2016 1 commit
  3. 17 Apr, 2015 6 commits
  4. 07 May, 2014 1 commit
    • Sven Zimmer's avatar
      minor refactoring · f73a5978
      Sven Zimmer authored
      first step for joining duplicate code from AnnoViewer with code from
      DeptreeViewer: DepTreeDisplay now abstract class, former DepTreeDisplay
      now called StandardDisplay
  5. 05 May, 2014 1 commit
  6. 30 Jun, 2013 1 commit
  7. 20 Sep, 2012 1 commit
  8. 28 Jul, 2012 1 commit
    • Niels Beuck's avatar
      additional virtual guards · 677e7fdd
      Niels Beuck authored
      new alternative usage of nonspec as head of fragments (there should probably be a way to switch without changing the grammar file)
  9. 18 Jun, 2012 1 commit
  10. 31 May, 2012 1 commit
  11. 30 May, 2012 1 commit
    • Arne Köhn's avatar
      Add two fixes for Malt constraints · 24dcc668
      Arne Köhn authored
       - maltlabel was checking the wrong prediction for existence
       - 'malt prediziert in die Zukunft' is not applicable for virtual nodes
  12. 25 May, 2012 3 commits
    • Arne Köhn's avatar
      Use the Maltparser as a predictor · 005fbb3a
      Arne Köhn authored
      This patch introduces the new MaltPredictor which uses Maltparser
      to predict the regent for each word. For this it changes the prediction
      interface so that any AVM can be used as prediction value, not only
      Two new predicates are introduced:
      notanumber(X) checks if X is not an AVMNumber.
      This is needed to check whether a prediction was set or the default value
      (AVMNumber(1.0)) is returned.
      isLexemeOf(l,g) checks if a given lexeme l is the lexeme of the grapheme g.
      Since Maltparser is unable to perform lexeme disambiguation, we can only
      predict graphemes while the grammar works on lexemes. This predicate allows
      to bridge this gap.
    • Niels Beuck's avatar
      another hotfix for the grammar... · b22ce034
      Niels Beuck authored
    • Niels Beuck's avatar
  13. 24 May, 2012 1 commit
  14. 12 Apr, 2012 1 commit
    • Arne Köhn's avatar
      separate lexicon and grammar · 2b10215a
      Arne Köhn authored
      Instead of inputSeq there is now cdggrammar and lexicon. This way,
      no lexicon entry can get lost in the grammar (and vice versa).
      Additionally, all lexicon entries from the grammar are deleted;
      those need to be in the lexicon.
  15. 26 Mar, 2012 1 commit
    • Niels Beuck's avatar
      Incremental parsing now starts with the empty structure. · 5c32acdd
      Niels Beuck authored
      The initial analysis in incremental parsing will now be the one with 0 words instead of the one for the first chunk.
      Thus, no special handling for the first chunk in incremental parsing is necessary any more.
      The virtual node constellation, (e.g. predicting verb and subject) for this initial structure can be set in the config.
  16. 02 Mar, 2012 1 commit