Search Descriptions

Main Topics

Search Publications


author

title

other

year

Parse Quality

Syntax based models are trained on parallel corpora that have been automatically annotated with syntactic parsers. The quality of the parsers obviously matters.

Parse Quality is the main subject of 7 publications.

Publications

The amount of training data for statistical parsers is one factor for the quality of parsers. Quirk and Corston-Oliver (2006) show that parsers trained on relatively small treebanks of only 250 sentences give decent performance, and while there are gains with larger treebanks, they expect these to be diminishing with treebanks larger than the ones currently available, i.e. over 40,000 sentences.

Benchmarks

Discussion

Related Topics

New Publications

  • Morishita et al. (2015)
  • Mi et al. (2008)
  • Goto et al. (2011)
  • Liu et al. (2012)
  • Neubig et al. (2012)
  • Popel et al. (2011)

Actions

Download

Contribute