--------------------------------  GRID search  --------------------------------

1. To calculate the similarities of rules (Jaccard calculation) run

		./SAFRAN_win.exe calcjacc jacc.txt

	for Windows or

		./SAFRAN_unix calcjacc jacc.txt

	for Unix systems.

2. To recreate the clustering with grid search run 

		./SAFRAN_win.exe learnnrnoisy learnnrnoisy_grid.txt

	for Windows or

		./SAFRAN_unix calcjacc learnnrnoisy learnnrnoisy_grid.txt

	for Unix systems.

3. To apply the cluster on the test set of FB15K-237 run

		./SAFRAN_win.exe applynrnoisy applynrnoisy_grid.txt

	for Windows or

		./SAFRAN_unix applynrnoisy applynrnoisy_grid.txt

	for Unix systems.

4. For evaluation run
		
		python eval.py ./FB15K-237/predictions/grid ./FB15K-237/test.txt


-------------------------------  RANDOM search  -------------------------------

1. If you have previously ran 1. of GRID search you can omit this step, otherwise proceed with the instructions in 1. of GRID search.

2. To recreate the clustering with random search run 

		./SAFRAN_win.exe learnnrnoisy learnnrnoisy_random.txt

	for Windows or

		./SAFRAN_unix calcjacc learnnrnoisy learnnrnoisy_random.txt

	for Unix systems.

3. To apply the cluster on the test set of FB15K-237 run

		./SAFRAN_win.exe applynrnoisy applynrnoisy_random.txt

	for Windows or

		./SAFRAN_unix applynrnoisy applynrnoisy_random.txt

	for Unix systems.


4. For evaluation run
		
		python eval.py ./FB15K-237/predictions/random ./FB15K-237/test.txt
