def evaluate_our_method(website, iter_count, data_count, attibute=["profession"]):
params = [
{"kernel": "rbf", "C": 100, "gamma": 0.001},
{"kernel": "rbf", "C": 10, "gamma": 0.001},
{"kernel": "rbf", "C": 10, "gamma": 0.001},
]
print("======Our; Iter Count: %d======" % iter_count)
embedding = get_neibor_embedding("./embedding/%s_user_embedding_using_neibors_%d.data.json" % (website, iter_count))
uids = [line.strip() for line in open("./%s_intersect_uid.data" % website)]
# evaluate(get_label(website, 1, gender_reg), embedding)
# evaluate(get_label(website, 2, age_reg), embedding)
# evaluate(get_label(website, 3, location_reg), embedding)
if "gender" in attibute:
results = simple_evaluate(
get_label(website, 1, gender_reg),
embedding,
params[0],
data_count,
data_range=total_data_arange["%s_gender" % (website)],
uids=uids,
)
output_results(results, "%s_%s_%s_result_%d.out" % (website, "gender", "our", iter_count))
if "age" in attibute:
results = simple_evaluate(
get_label(website, 2, age_reg),
embedding,
params[1],
data_count,
data_range=list(arange(0.05, 0.95, 0.05)),
uids=uids,
)
output_results(results, "%s_%s_%s_result_%d.out" % (website, "age", "our", iter_count))
if "location" in attibute:
results = simple_evaluate(
get_label(website, 3, location_reg),
embedding,
params[2],
data_count,
data_range=total_data_arange["%s_location" % (website)],
uids=uids,
)
output_results(results, "%s_%s_%s_result_%d.out" % (website, "location", "our", iter_count))
if "profession" in attibute:
results = simple_evaluate(
get_label(website, 4, profession_reg),
embedding,
params[2],
data_count,
data_range=total_data_arange["%s_profession" % (website)],
uids=uids,
)
output_results(results, "%s_%s_%s_result_%d.out" % (website, "profession", "our", iter_count))
def evaluate_baseline(fname, data_count):
deepwalk_params = [
{'kernel': 'rbf',
'C': 1000,
'gamma': 0.001}, {'kernel': 'rbf',
'C': 100,
'gamma': 0.001}, {'kernel': 'rbf',
'C': 10,
'gamma': 0.001}
]
line_params = [
{'kernel': 'linear', 'C': 1},
{'kernel': 'rbf', 'C': 100, 'gamma': 0.001},
{'kernel': 'rbf', 'C': 1000, 'gamma': 0.001}
]
if 'deepwalk' in fname:
print('======Deepwalk======')
params = deepwalk_params
if 'line' in fname:
print('======LINE======')
params = line_params
if 'weibo' in fname:
website = 'weibo'
else:
website = 'zhihu'
embedding = get_simple_embedding(fname)
# evaluate(get_label(website, 1, gender_reg), embedding)
# evaluate(get_label(website, 2, age_reg), embedding)
# evaluate(get_label(website, 3, location_reg), embedding)
simple_evaluate(
get_label(website, 1, gender_reg), embedding, params[0], data_count)
def evaluate_our_method(website, iter_count, data_count):
params = [
{'kernel': 'rbf',
'C': 100,
'gamma': 0.001}, {'kernel': 'rbf',
'C': 10,
'gamma': 0.001}, {'kernel': 'rbf',
'C': 10,
'gamma': 0.001}
]
print('======Our; Iter Count: %d======' % iter_count)
embedding = get_neibor_embedding(
'./embedding/%s_user_embedding_using_neibors_%d.data.json' %
(website, iter_count)
)
# evaluate(get_label(website, 1, gender_reg), embedding)
# evaluate(get_label(website, 2, age_reg), embedding)
# evaluate(get_label(website, 3, location_reg), embedding)
simple_evaluate(
get_label(website, 1, gender_reg), embedding, params[0], data_count)
# simple_evaluate(get_label(website, 2, age_reg), embedding, params[1])
simple_evaluate(
get_label(website, 3, location_reg), embedding, params[2], data_count)
def main():
website = "outputacm"
method = "line"
count = 100
if method == "line":
embedding = get_simple_embedding("./embedding/%s_user_embedding_using_line.data.json" % website)
if method == "our":
embedding = get_neibor_embedding("./embedding/%s_user_embedding_using_neibors_%d.data.json" % (website, count))
print(len(embedding))
results = simple_evaluate(
get_label(website, 1, outputacm_reg),
embedding,
data_range=list(arange(0.1, 1, 0.1)),
uid_file="./%s_intersect_uid.data" % website,
)
output_results(results, "%s_%s_result_%d.out" % (website, method, count))
pass
def evaluate_baseline(fname, data_count, attibute=["profession"]):
deepwalk_params = [
{"kernel": "rbf", "C": 1000, "gamma": 0.001},
{"kernel": "rbf", "C": 100, "gamma": 0.001},
{"kernel": "rbf", "C": 10, "gamma": 0.001},
]
line_params = [
{"kernel": "linear", "C": 1},
{"kernel": "rbf", "C": 100, "gamma": 0.001},
{"kernel": "linear", "C": 1000, "gamma": 0.001},
]
if "deepwalk" in fname:
print("======Deepwalk======")
baseline = "deepwalk"
params = deepwalk_params
if "line" in fname:
print("======LINE======")
params = line_params
baseline = "line"
if "weibo" in fname:
website = "weibo"
else:
website = "zhihu"
embedding = get_simple_embedding(fname)
uids = [line.strip() for line in open("./%s_intersect_uid.data" % website)]
# evaluate(get_label(website, 1, gender_reg), embedding)
# evaluate(get_label(website, 2, age_reg), embedding)
# evaluate(get_label(website, 3, location_reg), embedding)
if "gender" in attibute:
results = simple_evaluate(
get_label(website, 1, gender_reg),
embedding,
params[0],
data_count,
data_range=total_data_arange["%s_gender" % (website)],
uids=uids,
)
output_results(results, "%s_%s_%s_result.out" % (website, "gender", baseline))
if "age" in attibute:
results = simple_evaluate(
get_label(website, 2, age_reg),
embedding,
params[1],
data_count,
data_range=list(arange(0.05, 0.95, 0.05)),
uids=uids,
)
output_results(results, "%s_%s_%s_result.out" % (website, "age", baseline))
if "location" in attibute:
results = simple_evaluate(
get_label(website, 3, location_reg),
embedding,
params[2],
data_count,
data_range=total_data_arange["%s_location" % (website)],
uids=uids,
)
output_results(results, "%s_%s_%s_result.out" % (website, "location", baseline))
if "profession" in attibute:
results = simple_evaluate(
get_label(website, 4, profession_reg),
embedding,
params[2],
data_count,
data_range=total_data_arange["%s_profession" % (website)],
uids=uids,
)
output_results(results, "%s_%s_%s_result.out" % (website, "profession", baseline))
