how fast is CKY parsing on a Treebank grammar?

[jeisner]

Should convert the Penn Treebank to Dyna rules (by running an external script manually or as a filter, see #7). Then count and normalize to get the grammar. Finally, try parsing some of the training sentences or some held-out sentences.

[timvieira]

How hard would it be to do tree binarization in Dyna? It's a little awkward without nice Dyna lists syntax and operators and backward chaining.

[jeisner]

How long till list syntax arrives in Dyna 2? It would be helpful to have for the class, I bet. But for now, you can write the program with list syntax first and then munge it to use explicit cons. Or I can just give you a pre-binarized version of the Treebank.