Parsers

  • Some software links below are temporarily unavailable. If you need something right away, please feel free to contact me.
  • LRDep: a shift-reduce dependency parser that processes input from left-to-right and produces a rightmost derivation.
  • GDep (GENIA Dependency parser): a dependency parser for biomedical text.
  • GRASP: a parser for transcripts of child language in the CHILDES database.
  • CLEDep: A very simple Perl implementation of the MST parser of McDonald et al. (2005) using the Chu-Liu/Edmonds algorithm and a simple structured perceptron. Nothing fancy here, but useful if you are just getting started and want to find out more about how MST dependency parsing works with discriminative online learning.
  • SimpleDep: A simple greedy transition based parser. It is a little-known fact that the Yamada & Matsumoto 2003 dependency parsing algorithm differs from what is widely referred to as arc standard parsing by only one line of code. This is an implementation of arc standard dependency parsing in the style of the Yamada & Matsumoto description of the algorithm. The difference is in the line $i -= 2, which makes the Y&M algorithm exactly the same as the shift-reduce arc standard algorithm. (A python implementation by Michael Heilman is also available.)
  • An RST discourse parser based on my IWPT 2009 paper (but modified and updated in a few ways) is available, thanks in large part to Michael Heilman, with contributions from others at ETS.

  • The code for the Dynamic Programming shift-reduce parser in Huang and Sagae 2010 is available from my colleague Liang Huang, who wrote the code. (My perl scripts for conversion between Liang's dependency format and CoNLL format: lh2conll.pl and conll2lh.pl)

Data

The child language transcripts with syntactic annotations are available directly from the CHILDES Database site. (in the main CHILDES site you will also find links to downloadable versions of the CHILDES treebank.)