-
Notifications
You must be signed in to change notification settings - Fork 0
/
train.py
72 lines (60 loc) · 3.77 KB
/
train.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
import subprocess
import os
from openbiolink.obl2021 import OBL2021Dataset
def main():
working_dir = os.path.abspath('obl2021').replace("\\","/")
print(f"Working directory is {working_dir}")
dl = OBL2021Dataset()
# Learn rules with AnyBURL
with open(os.path.join(working_dir, "learn_.txt"),"w") as outfile:
outfile.write(f"PATH_TRAINING = {working_dir}/train.tsv" + "\n")
outfile.write(f"PATH_OUTPUT = {working_dir}/rules" + "\n")
outfile.write(f"SNAPSHOTS_AT = 1000" + "\n")
outfile.write(f"WORKER_THREADS = 22" + "\n")
subprocess.call([r"java", "-cp", "./AnyBURL-RE/AnyBURL-RE.jar", "de.unima.ki.anyburl.LearnReinforced", os.path.join(working_dir, "learn_.txt")])
# Apply rules (Maximum aggregation)
with open(os.path.join(working_dir, "apply_.txt"),"w") as outfile:
outfile.write(f"PATH_TRAINING = {working_dir}/train.tsv" + "\n")
outfile.write(f"PATH_TEST = {working_dir}/test.tsv" + "\n")
outfile.write(f"PATH_VALID = {working_dir}/valid.tsv" + "\n")
outfile.write(f"PATH_RULES = {working_dir}/rules-1000" + "\n")
outfile.write(f"PATH_OUTPUT = {working_dir}/prediction_max" + "\n")
outfile.write(f"TOP_K_OUTPUT = 10" + "\n")
subprocess.call([r"./SAFRAN/build/Release/SAFRAN", "applymax", os.path.join(working_dir, "apply_.txt")])
# Calculate Jaccard indices
with open(os.path.join(working_dir, "jacc_.txt"),"w") as outfile:
outfile.write(f"PATH_TRAINING = {working_dir}/train.tsv" + "\n")
outfile.write(f"PATH_TEST = {working_dir}/test.tsv" + "\n")
outfile.write(f"PATH_VALID = {working_dir}/valid.tsv" + "\n")
outfile.write(f"PATH_RULES = {working_dir}/rules-1000" + "\n")
outfile.write(f"PATH_JACCARD = {working_dir}" + "\n")
outfile.write(f"RESOLUTION = 200" + "\n")
subprocess.call([r"./SAFRAN/build/Release/SAFRAN", "calcjacc", os.path.join(working_dir, "jacc_.txt")])
# Learn NRNO cluster (random search)
with open(os.path.join(working_dir, "learn_.txt"),"w") as outfile:
outfile.write(f"PATH_TRAINING = {working_dir}/train.tsv" + "\n")
outfile.write(f"PATH_TEST = {working_dir}/test.tsv" + "\n")
outfile.write(f"PATH_VALID = {working_dir}/valid.tsv" + "\n")
outfile.write(f"PATH_RULES = {working_dir}/rules-1000" + "\n")
outfile.write(f"PATH_JACCARD = {working_dir}" + "\n")
outfile.write(f"DISCRIMINATION_BOUND = 1000" + "\n")
outfile.write(f"TOP_K_OUTPUT = 50" + "\n")
outfile.write(f"PATH_CLUSTER = {working_dir}/cluster_nrno" + "\n")
outfile.write(f"STRATEGY = random" + "\n")
outfile.write(f"ITERATIONS = 10000" + "\n")
outfile.write(f"RESOLUTION = 10" + "\n")
outfile.write(f"SEED = 0" + "\n")
subprocess.call([r"./SAFRAN/build/Release/SAFRAN", "learnnrnoisy", os.path.join(working_dir, "learn_.txt")])
# Apply NRNO (random search)
with open(os.path.join(working_dir, "apply_.txt"),"w") as outfile:
outfile.write(f"PATH_TRAINING = {working_dir}/train.tsv" + "\n")
outfile.write(f"PATH_TEST = {working_dir}/test.tsv" + "\n")
outfile.write(f"PATH_VALID = {working_dir}/valid.tsv" + "\n")
outfile.write(f"PATH_RULES = {working_dir}/rules-1000" + "\n")
outfile.write(f"PATH_CLUSTER = {working_dir}/cluster_nrno" + "\n")
outfile.write(f"PATH_OUTPUT = {working_dir}/prediction_nrno" + "\n")
outfile.write(f"WORKER_THREADS = -1" + "\n")
outfile.write(f"TOP_K_OUTPUT = 10" + "\n")
subprocess.call([r"./SAFRAN/build/Release/SAFRAN", "applynrnoisy", os.path.join(working_dir, "apply_.txt")])
if __name__ == "__main__":
main()