To perform a sample run, do:

./src/classify.py wsj01.mrg "PARSER".mrg > "PARSER".errors

We recommend viewing like this:

less -x3 "PARSER".errors

The output contains colours that are not rendered correctly in most text
editors.

In this directory we have included parses for the first ten sentences in
section 01 of the Penn Treebank.  The files correspond to:

berkeley.mrg - Berkeley Parser
bikel.mrg - Bikel Parser
bubs.mrg - BUBS Parser
charniak_basic.mrg - Charniak Parser, 1-best
charniak_reranking.mrg - Charniak Parser, with reranker
charniak_selftrained_basic.mrg - Charniak Parser, 1-best with the self-trained model
charniak_selftrained_reranking.mrg - Charniak Parser, with reranker and self-trained model
collins.1.mrg - Collins Model 1
collins.2.mrg - Collins Model 2
collins.3.mrg - Collins Model 3
henderson.mrg - SSN Parser
stanford.mrg - Stanford Unlexicalised Parser
stanford_factored.mrg - Stanford Lexicalised Parser
(Double DOP is not included as we only just got the parser working in time)

wsj01.mrg - Gold trees
wsj01.tok - Just text
