COLING22_CTM

Experiments codes for the paper:

Dugang Liu, Weihao Du, Lei Li, Weike Pan and Zhong Ming. Augmenting Legal Judgment Prediction with Contrastive Case Relations. To appear in COLING '22.

Please cite our COLING '22 paper if you use our codes. Thanks!

---

Usage

The execution process of the main experiment:

• Switch to working directory:

cd COLING22_CTM/CTM_small

cd COLING22_CTM/CTM_big

• Carry out data exploration and obtain experimental data:

python3 data_exploration.py

python3 get_processed_data.py
python3 generate_data_structure.py

python3 get_processed_data.py -d 'big/'
python3 generate_data_structure.py -d 'big/'

• Tuning model (searcher='optuna'):

  CUDA_VISIBLE_DEVICES=0 nohup python3 -u tune_parameters.py -tb 'ctm_tuning_0.csv' -y 'config/ctm.yml' -s 0 >ctm_out_0 2>&1 &
  
  

• Train the model according to the best parameters, save the parameters, and output the results:

  CUDA_VISIBLE_DEVICES=0 python3 reproduce_paper_results.py