Here you will find some material on Mutual Information and Relative Entropy based parsing experiments.

The Bibtex and Latex files contain a more detailed description and documentation of the project. Some results are in the spreadsheet file, in particular the incremental parsing of the Penn Treebank sentences etc.

The code-base will be extended soon, an implementation in Scheme is the next one, using various other statistical metrics and distance measures.

Back to Damir Cavar’s homepage...