def read_logmap_mapping(filename):
mappings_dict = dict()
with open(filename) as f:
for line in f.readlines():
m = line.strip().split('|')
left_c = uri_prefix(m[0])
right_c = uri_prefix(m[1])
if left_c in mappings_dict:
mappings_dict[left_c].append(right_c)
else:
mappings_dict[left_c] = [right_c]
return mappings_dict
def append_super_class(c, p):
p.append(uri_prefix(uri=c.iri))
supclasses = super_classes(c=c)
if owl.Thing in supclasses or len(supclasses) == 0 or supclasses is None:
return p
else:
return append_super_class(c=supclasses[0], p=p)
def get_class_name(o):
c_name = dict()
for c in o.classes():
name = c.name
labels = c.label.en + c.label
names = [name, labels[0]] if len(labels) > 0 else [name, None]
c_name[uri_prefix(uri=c.iri)] = names
return c_name
def read_oaei_mappings(file_name):
tree = ET.parse(file_name)
mappings_str = list()
all_mappings_str = list()
for t in tree.getroot().getchildren():
for m in t.getchildren():
if 'map' in m.tag:
for c in m.getchildren():
mapping = list()
mv = '?'
for i, v in enumerate(c.getchildren()):
if i < 2:
for value in v.attrib.values():
mapping.append(uri_prefix(value).lower())
break
if i == 3:
mv = v.text
all_mappings_str.append('|'.join(mapping))
if not mv == '?':
mappings_str.append('|'.join(mapping))
return mappings_str, all_mappings_str
left_paths = [
line.strip().split(',') for line in open(FLAGS.left_path_file).readlines()
]
right_paths = [
line.strip().split(',')
for line in open(FLAGS.right_path_file).readlines()
]
left_names = json.load(open(FLAGS.left_class_name_file))
right_names = json.load(open(FLAGS.right_class_name_file))
mappings, mappings_n = list(), list()
with open(FLAGS.candidate_file) as f:
for i, line in enumerate(f.readlines()):
m = line.strip().split(', ')[1] if ', ' in line else line.strip()
m_split = m.split('|')
c1 = uri_prefix(uri=m_split[0])
c2 = uri_prefix(uri=m_split[1])
n1 = get_label(cls=c1,
paths=left_paths,
names=left_names,
label_type='path',
keep_uri=(FLAGS.keep_uri == 'yes'))
n2 = get_label(cls=c2,
paths=right_paths,
names=right_names,
label_type='path',
keep_uri=(FLAGS.keep_uri == 'yes'))
origin = 'i=%d|%s|%s' % (i + 1, c1, c2)
name = '%s|%s' % (n1, n2)
mappings.append(origin)
all_mappings_str.append('|'.join(mapping))
if not mv == '?':
mappings_str.append('|'.join(mapping))
return mappings_str, all_mappings_str
if __name__ == "__main__":
ref_mappings_str, ref_all_mappings_str = read_oaei_mappings(file_name=FLAGS.oaei_GS_file)
ref_excluded_mappings_str = set(ref_all_mappings_str) - set(ref_mappings_str)
anchor_mappings_str = list()
with open(FLAGS.anchor_mapping_file) as f:
for line in f.readlines():
tmp = line.strip().split('|')
anchor_mappings_str.append('%s|%s' % (uri_prefix(tmp[0]).lower(), uri_prefix(tmp[1]).lower()))
pred_mappings_str = list()
with open(FLAGS.prediction_out_file) as f:
lines = f.readlines()
for j in range(0, len(lines), 3):
tmp = lines[j].split('|')
if float(tmp[3]) >= FLAGS.threshold:
pred_mappings_str.append('%s|%s' % (tmp[1].lower(), tmp[2].lower()))
for a in anchor_mappings_str:
if a not in pred_mappings_str:
pred_mappings_str.append(a)
recall_num = 0
for s in ref_mappings_str:
s_n = 0
for sample in samples:
tmp = sample.split('|')
if '%s|%s' % (tmp[0], tmp[1]) not in GS:
s_n += 1
if len(tmp) >= 4 and tmp[3] == 'true':
sv_n += 1
print(
'All three systems: sampled mappings not in GS: %d, correct samples: %d, sampled precision: %f'
% (s_n, sv_n, sv_n / s_n))
from lib.Label import uri_prefix
mappings = set()
for line in open(LogMap_output_mapping_file).readlines():
tmp = line.strip().split('|')
c1 = uri_prefix(uri=tmp[0])
c2 = uri_prefix(uri=tmp[1])
mappings.add('%s|%s' % (c2, c2))
for line in open(LogMap_ML_output_mapping_file).readlines():
mappings.add(line.strip())
for line in open(AML_output_mapping_file).readlines():
mappings.add(line.strip())
G_M_n = len(mappings - GS)
print('mappings out of GS: %d' % G_M_n)
num = len(GS) + sv_n / s_n * G_M_n
print('approximate GS size: %d' % num)
print('approximate recall: %.3f' % (app_TP / num))
left_paths = [
line.strip().split(',')
for line in open(FLAGS.left_path_file).readlines()
]
right_paths = [
line.strip().split(',')
for line in open(FLAGS.right_path_file).readlines()
]
mappings = list()
rule_violated_mappings = list()
with open(FLAGS.anchor_mapping_file) as f:
for i, line in enumerate(f.readlines()):
tmp = line.strip().split(', ')[1] if ', ' in line else line.strip()
tmp2 = tmp.split('|')
c1 = uri_prefix(uri=tmp2[0])
c2 = uri_prefix(uri=tmp2[1])
n1 = get_label(cls=c1,
paths=left_paths,
names=left_names,
label_type='path',
keep_uri=(FLAGS.keep_uri == 'yes'))
n2 = get_label(cls=c2,
paths=right_paths,
names=right_names,
label_type='path',
keep_uri=(FLAGS.keep_uri == 'yes'))
if not n1 == '""' and not n2 == '""':
if violate_rules(p1_str=n1,
p2_str=n2,
