if __name__ == "__main__":
FEATURE_SIZE = 105
NUM_INTERACTION = 2000000
click_models = ["informational", "navigational", "perfect"]
# click_models = ["perfect"]
Learning_rate = 0.1
num_groups = 4
dataset_path = "datasets/clueweb09_intent_change.txt"
intent_path = "intents"
output_fold = "results/SDBN/PDGD/group_mixed_2m"
train_set = LetorDataset(dataset_path,
FEATURE_SIZE,
query_level_norm=True,
binary_label=True)
intent_paths = [
"{}/1.txt".format(intent_path), "{}/2.txt".format(intent_path),
"{}/3.txt".format(intent_path), "{}/4.txt".format(intent_path)
]
for click_model in click_models:
mp.Process(target=job,
args=(click_model, Learning_rate, NUM_INTERACTION, 1,
train_set, intent_paths, output_fold,
num_groups)).start()
.format(f, model_type, 2, r), "wb") as fp:
pickle.dump(cndcg_scores2, fp)
with open(
"../results/reduction/mq2007/PDGD/fold{}/{}_ranker{}_run{}_final_weight.txt"
.format(f, model_type, 2, r), "wb") as fp:
pickle.dump(final_weight2, fp)
print("PDGD tau{} fold{} {} run{} finished!".format(
tau, f, model_type, r))
if __name__ == "__main__":
FEATURE_SIZE = 46
NUM_INTERACTION = 10000
# click_models = ["informational", "navigational", "perfect"]
click_model = "informational"
Learning_rate = 0.1
dataset_fold = "../datasets/2007_mq_dataset"
output_fold = "mq2007"
# taus = [0.1, 0.5, 1.0, 5.0, 10.0]
tau = 1
# for 5 folds
for f in range(1, 6):
training_path = "{}/Fold{}/train.txt".format(dataset_fold, f)
test_path = "{}/Fold{}/test.txt".format(dataset_fold, f)
train_set = LetorDataset(training_path, FEATURE_SIZE)
test_set = LetorDataset(test_path, FEATURE_SIZE)
mp.Process(target=job,
args=(click_model, f, train_set, test_set, tau)).start()
if click_model == "perfect":
pc = [0.0, 0.2, 0.4, 0.8, 1.0]
ps = [0.0, 0.0, 0.0, 0.0, 0.0]
elif click_model == "navigational":
pc = [0.05, 0.3, 0.5, 0.7, 0.95]
ps = [0.2, 0.3, 0.5, 0.7, 0.9]
elif click_model == "informational":
pc = [0.4, 0.6, 0.7, 0.8, 0.9]
ps = [0.1, 0.2, 0.3, 0.4, 0.5]
for f in fold_range:
training_path = "{}/Fold{}/train.txt".format(
dataset_fold, f)
test_path = "{}/Fold{}/test.txt".format(dataset_fold, f)
train_set = LetorDataset(training_path,
FEATURE_SIZE,
query_level_norm=norm,
cache_root="../datasets/cache")
test_set = LetorDataset(test_path,
FEATURE_SIZE,
query_level_norm=norm,
cache_root="../datasets/cache")
print(dataset_fold, click_model, f, batch_size)
p = mp.Process(target=job,
args=(click_model, f, train_set, test_set,
output_fold, batch_size, pc, ps))
p.start()
processors.append(p)
for p in processors:
p.join()
if cm.name == "Mixed":
line += s_name + " "
line += "\n"
f.write(line)
# if index % 10000 == 0:
# print("write %d/%d queries" % (index, num_queries))
f.close()
print(cm.name, "unseen_set finished!")
# %%
if __name__ == "__main__":
# %%
train_path = "../datasets/ltrc_yahoo/set1.train.txt"
print("loading training set.......")
train_set = LetorDataset(train_path, 700)
# %%
# pc = [0.4, 0.6, 0.7, 0.8, 0.9]
# ps = [0.1, 0.2, 0.3, 0.4, 0.5]
pc = [0.05, 0.3, 0.5, 0.7, 0.95]
ps = [0.2, 0.3, 0.5, 0.7, 0.9]
# click_models = [DCTR(pc), SDBN(pc, ps), UBM(pc), SDBN_reverse(pc, ps)]
# Mixed_model = Mixed(click_models)
# simulators = [DCTR(pc), SDBN(pc, ps), UBM(pc), SDBN_reverse(pc, ps), Mixed_model]
simulators = [SDBN_reverse(pc, ps)]
for id in range(2, 16):
pool = []
for cm in simulators:
NUM_INTERACTION = 100000
click_models = ["informational", "perfect"]
# click_models = ["perfect"]
# dataset_fold = "../datasets/MSLR10K"
# dataset_fold = "../datasets/2007_mq_dataset"
output_fold = "results/istella/DBGD"
# output_fold = "results/mslr10k/DBGD"
# output_fold = "results/yahoo/PMGD"
# taus = [0.1, 0.5, 1.0, 5.0, 10.0]
alpha = 0.01
delta = 1
num_rankers = 1
# for 5 folds
for f in range(1, 2):
# training_path = "{}/Fold{}/train.txt".format(dataset_fold, f)
# test_path = "{}/Fold{}/test.txt".format(dataset_fold, f)
training_path = "../datasets/istella/train.txt"
test_path = "../datasets/istella/test.txt"
print("loading dataset.....")
# training_path = "../datasets/ltrc_yahoo/set1.train.txt"
# test_path = "../datasets/ltrc_yahoo/set1.test.txt"
train_set = LetorDataset(training_path, FEATURE_SIZE, query_level_norm=True)
test_set = LetorDataset(test_path, FEATURE_SIZE, query_level_norm=True)
# for 3 click_models
for click_model in click_models:
p = mp.Process(target=job, args=(click_model, f, train_set, test_set,
delta, alpha, FEATURE_SIZE, num_rankers, output_fold))
p.start()
Learning_rate = 0.1
dataset_fold = "../datasets/2007_mq_dataset"
output_fold = "mq2007"
tau = 1.0
# for 5 folds
# for f in range(1, 6):
# training_path = "{}/Fold{}/train.txt".format(dataset_fold, f)
# test_path = "{}/Fold{}/test.txt".format(dataset_fold, f)
# train_set = LetorDataset(training_path, FEATURE_SIZE)
# test_set = LetorDataset(test_path, FEATURE_SIZE)
#
# # for 3 click_models
# for click_model in click_models:
# mp.Process(target=job, args=(click_model, f, train_set, test_set, tau)).start()
training_path = "{}/Fold{}/train.txt".format(dataset_fold, 1)
train_set = LetorDataset(training_path, FEATURE_SIZE)
lsh = LSHash(2, FEATURE_SIZE, num_hashtables=2)
ranker1 = []
ranker2 = []
ranker3 = []
ranker4 = []
print(len(train_set.get_all_querys()))
for q in train_set.get_all_querys():
query = np.mean(train_set.get_all_features_by_query(q), axis=0)
code = lsh._hash(lsh.uniform_planes[0], query)
print(code)
print(lsh._hash(lsh.uniform_planes[1], query))
print()
if code == '00':
ranker1.append(q)
elif code == '01':
exist_ok=True) # create directory if not exist
with open(
"{}/fold{}/{}_run{}_ndcg.txt".format(output_fold, f, model_type, r),
"wb") as fp:
pickle.dump(ndcg_scores, fp)
with open(
"{}/fold{}/{}_run{}_cndcg.txt".format(output_fold, f, model_type, r),
"wb") as fp:
pickle.dump(cndcg_scores, fp)
with open(
"{}/fold{}/{}_run{}_weights.txt".format(output_fold, f, model_type, r),
"wb") as fp:
pickle.dump(final_weights, fp)
print("PDGD fold{} {} run{} finished!".format(f, model_type, r))
if __name__ == "__main__":
# click_models = ["informational", "navigational", "perfect"]
click_models = ["perfect", "informational"]
dataset_fold = "../datasets/MSLR10k"
output_fold = "../results/exploration/PDGD/MSLR10K/random"
# for 5 folds
for f in range(1, 2):
training_path = "{}/Fold{}/train.txt".format(dataset_fold, f)
test_path = "{}/Fold{}/test.txt".format(dataset_fold, f)
train_set = LetorDataset(training_path, FEATURE_SIZE, query_level_norm=True, cache_root="../datasets/cache", binary_label=3)
test_set = LetorDataset(test_path, FEATURE_SIZE, query_level_norm=True, cache_root="../datasets/cache", binary_label=3)
for click_model in click_models:
mp.Process(target=job, args=(click_model, f, train_set, test_set, output_fold)).start()