how-to-run-the-experiments.md

Files to run for Experiments

The following is a description on how to run the experiments.

The experiments investigate how the confidence estimators vary i.f.o. varying propensity scores

The SCAR_p selection mechanism

Known propensity scores (Q1)

For a minimal working example, see: run_exp_min_working_ex.py or run_exp_min_working_ex_multiple_label_frequencies.py

To run the experiment for all predicates in yago3_10, run: run_yago3_10.py

Noisy propensity scores (Q2)

For a minimal working example, see: run_exp_min_working_ex.py

To run the experiment for all predicates in yago3_10, run: run_yago3_10.py

The SAR_{group} selection mechanism

Known propensity scores (Q1)

For a minimal working example, see: run_exp_min_working_ex.py or run_exp_min_working_ex_multiple_prop_scores.py

To run the experiments for yago3_10, run: run_yago3_10.py

Noisy propensity scores (Q2 and Q3)

For a minimal working example, see: run_exp_min_working_ex.py To run the experiments for yago3_10, run: run_yago3_10.py

The SAR_{popularity} selection mechanism

To run the experiments fo yago3_10, run: run_yago3_10_noise_as_fraction_of_log_growth_rate.py