ICLR Discourse Dataset

You will need to have downloaded Stanford CoreNLP and set $CORENLP_HOME to point to the unzipped directory.

To set up the environment and build the dataset, run

bash setup.sh

This step takes about 30 minutes, and should produce dataset files in the directory review_classification.