QA4PC Dataset (paper: Cross-Policy Compliance Detection via Question Answering)
Train Sets
To create training set or entailment and QA tasks, download and convert the ShARC data using the following commands:
wget https://sharc-data.github.io/data/sharc1-official.zip
unzip sharc1-official.zip
python create_train_from_sharc.py -sharc_dev_path sharc1-official/json/sharc_dev.json -sharc_train_path sharc1-official/json/sharc_train.json
Evaluation Sets
Entailment Data
The following files contain the data for the entailment task. This includes the policy + questions, a scenario and an answer (Yes, No, Maybe). Each datapoint also contain the information from the ShARC dataset such as tree_id and source_url.
- dev_entailment_qa4pc.json
- test_entailment_qa4pc.json
QA Data
The following files contain the data for the QA task.
- dev_sc_qa4pc.json
- test_sc_qa4pc.json
The following file contains the expression tree data for the dev and test sets. Each tree includes a policy, a set of questions and a logical expression.
- trees_dev_test_qa4pc.json