Read the paper; listen to the NLP Highlights Podcast
Download SARA.tar.gz
The statutes are located under statutes/source/ and the cases under cases/. The cases are split into train and test, following splits/{train,test}. This Github repo contains instructions and scripts to format the dataset into .txt files, typically used for machine reading and to run experiments.
Complementary statistics to the dataset can be found here.
To run the Prolog program, you need SWI-Prolog version 7.2.3 for amd64. Run prolog statutes/prolog/init.pl. This will load into memory the entirety of the Prolog version of the statutes, and open an interactive Prolog shell. From there, you can enter facts and call predicates.
Each case can be run individually, via e.g. prolog cases/s3306_a_2_B_neg.pl. The final :- halt statement ensures that no interactive shell remains open. If you want the shell to remain open, e.g. to try out other predicates with the facts of the case, comment the halt statement out.
You may run all the cases one after the other with bash code/run_cases_prolog.sh. Running this command is a good initial check to see whether the Prolog program is being interpreted and run correctly.