def test_mismatch(self):
self.assertAlmostEqual(
name_similarity(('Robin K.', 'Ryder'), ('Robin J.', 'Ryder')), 0)
self.assertAlmostEqual(
name_similarity(('Claire', 'Mathieu'), ('Claire', 'Kenyon-Mathieu')), 0)
self.assertAlmostEqual(
name_similarity(('Amanda P.', 'Brown'), ('Patrick', 'Brown')), 0)
def test_mismatch(self):
self.assertAlmostEqual(
name_similarity(('Robin K.', 'Ryder'), ('Robin J.', 'Ryder')), 0)
self.assertAlmostEqual(
name_similarity(('Claire', 'Mathieu'),
('Claire', 'Kenyon-Mathieu')), 0)
self.assertAlmostEqual(
name_similarity(('Amanda P.', 'Brown'), ('Patrick', 'Brown')), 0)
def test_symmetric(self):
pairs = [
(('Robin', 'Ryder'), ('Robin', 'Ryder')),
(('Robin', 'Ryder'), ('R.', 'Ryder')),
(('R.', 'Ryder'), ('R.', 'Ryder')),
(('Robin J.', 'Ryder'), ('R.', 'Ryder')),
(('Robin J.', 'Ryder'), ('R. J.', 'Ryder')),
(('R. J.', 'Ryder'), ('J.', 'Ryder')),
(('Robin', 'Ryder'), ('Robin J.', 'Ryder')),
(('W. Timothy', 'Gowers'), ('Timothy', 'Gowers')),
(('Robin K.', 'Ryder'), ('Robin J.', 'Ryder')),
(('Claire', 'Mathieu'), ('Claire', 'Kenyon-Mathieu')),
]
for a, b in pairs:
self.assertAlmostEqual(name_similarity(a, b), name_similarity(b, a))
def test_symmetric(self):
pairs = [
(('Robin', 'Ryder'), ('Robin', 'Ryder')),
(('Robin', 'Ryder'), ('R.', 'Ryder')),
(('R.', 'Ryder'), ('R.', 'Ryder')),
(('Robin J.', 'Ryder'), ('R.', 'Ryder')),
(('Robin J.', 'Ryder'), ('R. J.', 'Ryder')),
(('R. J.', 'Ryder'), ('J.', 'Ryder')),
(('Robin', 'Ryder'), ('Robin J.', 'Ryder')),
(('W. Timothy', 'Gowers'), ('Timothy', 'Gowers')),
(('Robin K.', 'Ryder'), ('Robin J.', 'Ryder')),
(('Claire', 'Mathieu'), ('Claire', 'Kenyon-Mathieu')),
]
for a, b in pairs:
self.assertAlmostEqual(name_similarity(a, b),
name_similarity(b, a))
def score(self, dataA, dataB):
# TODO: this score function is far from optimal
# refine it so that 'Claire Mathieu' and 'Claire Mathieu-Kenyon' gets
# a decent score
firstA, lastA = dataA
firstB, lastB = dataB
return name_similarity(dataA,dataB)
def get_or_create_by_orcid(cls, orcid, profile=None, user=None):
researcher = None
try:
researcher = Researcher.objects.get(orcid=orcid)
except Researcher.DoesNotExist:
if profile is None:
profile = OrcidProfile(id=orcid)
else:
profile = OrcidProfile(json=profile)
name = profile.name
homepage = profile.homepage
email = profile.email
researcher = Researcher.create_by_name(name[0],
name[1],
orcid=orcid,
user=user,
homepage=homepage,
email=email)
# Ensure that extra info is added.
save = False
for kw, val in [('homepage', homepage), ('orcid', orcid),
('email', email)]:
if not researcher.__dict__[kw] and val:
researcher.__dict__[kw] = val
save = True
if save:
researcher.save()
for variant in profile.other_names:
confidence = name_similarity(variant, variant)
name = Name.lookup_name(variant)
researcher.add_name_variant(name, confidence)
return researcher
def update_variants(self, reset=False):
"""
Sets the variants of this name to the candidates returned by variants_queryset
and which have a positive name similarity with the reference name.
.. todo::
This should rather rely on the name variants with confidence 1.0
"""
nvqs = self.namevariant_set.all()
if reset:
for nv in nvqs:
name = nv.name
nv.delete()
name.update_best_confidence()
current_name_variants = set()
else:
current_name_variants = set([nv.name_id for nv in nvqs])
last = self.name.last
for name in self.variants_queryset():
sim = name_similarity((name.first, name.last),
(self.name.first, self.name.last))
if sim > 0 and name.id not in current_name_variants:
self.add_name_variant(name, sim, force_update=reset)
def get_or_create_by_orcid(cls, orcid, profile=None, user=None):
researcher = None
try:
researcher = Researcher.objects.get(orcid=orcid)
except Researcher.DoesNotExist:
if profile is None:
profile = OrcidProfile(id=orcid)
else:
profile = OrcidProfile(json=profile)
name = profile.name
homepage = profile.homepage
email = profile.email
researcher = Researcher.create_by_name(name[0],name[1], orcid=orcid,
user=user, homepage=homepage, email=email)
# Ensure that extra info is added.
save = False
for kw, val in [('homepage',homepage),('orcid',orcid),('email',email)]:
if not researcher.__dict__[kw] and val:
researcher.__dict__[kw] = val
save = True
if save:
researcher.save()
for variant in profile.other_names:
confidence = name_similarity(variant, variant)
name = Name.lookup_name(variant)
researcher.add_name_variant(name, confidence)
return researcher
def update_variants(self, reset=False):
"""
Sets the variants of this name to the candidates returned by variants_queryset
and which have a positive name similarity with the reference name.
.. todo::
This should rather rely on the name variants with confidence 1.0
"""
nvqs = self.namevariant_set.all()
if reset:
for nv in nvqs:
name = nv.name
nv.delete()
name.update_best_confidence()
current_name_variants = set()
else:
current_name_variants = set([nv.name_id for nv in nvqs])
last = self.name.last
for name in self.variants_queryset():
sim = name_similarity((name.first,name.last),
(self.name.first,self.name.last))
if sim > 0 and name.id not in current_name_variants:
self.add_name_variant(name, sim, force_update=reset)
def update_variants(self):
"""
Sets the variants of this name to the candidates returned by variants_queryset
"""
for researcher in self.variants_queryset():
sim = name_similarity((researcher.name.first,researcher.name.last), (self.first,self.last))
if sim > 0:
old_sim = self.best_confidence
self.best_confidence = sim
if self.pk is None or old_sim < sim:
self.save()
NameVariant.objects.get_or_create(name=self,researcher=researcher,
defaults={'confidence':sim})
def score(self, dataA, dataB):
score = 0.
if dataA is None or dataB is None:
return 0.
for a in dataA:
for b in dataB:
firstA, lastA = a
firstB, lastB = b
score += name_similarity(a,b)
#score += name_tools.match(firstA+' '+lastA,firstB+' '+lastB)
# Previously, it was:
#if match_names(a,b):
# score += 1.
return score
def update_variants(self):
"""
Sets the variants of this name to the candidates returned by variants_queryset
"""
for researcher in self.variants_queryset():
sim = name_similarity(
(researcher.name.first, researcher.name.last),
(self.first, self.last))
if sim > 0:
old_sim = self.best_confidence
self.best_confidence = sim
if self.pk is None or old_sim < sim:
self.save()
NameVariant.objects.get_or_create(name=self,
researcher=researcher,
defaults={'confidence': sim})
def test_matching(self):
self.assertAlmostEqual(
name_similarity(('Robin', 'Ryder'), ('Robin', 'Ryder')), 0.8)
self.assertAlmostEqual(
name_similarity(('Robin', 'Ryder'), ('R.', 'Ryder')), 0.4)
self.assertAlmostEqual(
name_similarity(('R.', 'Ryder'), ('R.', 'Ryder')), 0.4)
self.assertAlmostEqual(
name_similarity(('Robin J.', 'Ryder'), ('R.', 'Ryder')), 0.3)
self.assertAlmostEqual(
name_similarity(('Robin J.', 'Ryder'), ('R. J.', 'Ryder')), 0.8)
self.assertAlmostEqual(
name_similarity(('R. J.', 'Ryder'), ('J.', 'Ryder')), 0.3)
self.assertAlmostEqual(
name_similarity(('Robin', 'Ryder'), ('Robin J.', 'Ryder')), 0.7)
def test_matching(self):
self.assertAlmostEqual(
name_similarity(('Robin', 'Ryder'), ('Robin', 'Ryder')), 0.8)
self.assertAlmostEqual(
name_similarity(('Robin', 'Ryder'), ('R.', 'Ryder')), 0.4)
self.assertAlmostEqual(
name_similarity(('R.', 'Ryder'), ('R.', 'Ryder')), 0.4)
self.assertAlmostEqual(
name_similarity(('Robin J.', 'Ryder'), ('R.', 'Ryder')), 0.3)
self.assertAlmostEqual(
name_similarity(('Robin J.', 'Ryder'), ('R. J.', 'Ryder')), 0.8)
self.assertAlmostEqual(
name_similarity(('R. J.', 'Ryder'), ('J.', 'Ryder')), 0.3)
self.assertAlmostEqual(
name_similarity(('Robin', 'Ryder'), ('Robin J.', 'Ryder')), 0.7)
def get_or_create_by_orcid(cls, orcid, profile=None, user=None):
"""
Creates (or returns an existing) researcher from its ORCID id.
:param profile: an :class:`OrcidProfile` object if it has already been fetched
from the API (otherwise we will fetch it ourselves)
:param user: an user to associate with the profile.
:returns: a :class:`Researcher` if everything went well, raises MetadataSourceException otherwise
"""
researcher = None
if orcid is None:
raise MetadataSourceException('Invalid ORCID id')
try:
researcher = Researcher.objects.get(orcid=orcid)
except Researcher.DoesNotExist:
if profile is None:
profile = OrcidProfile(id=orcid)
else:
profile = OrcidProfile(json=profile)
name = profile.name
homepage = profile.homepage
email = profile.email
researcher = Researcher.create_by_name(name[0],
name[1],
orcid=orcid,
user=user,
homepage=homepage,
email=email)
# Ensure that extra info is added.
save = False
for kw, val in [('homepage', homepage), ('orcid', orcid),
('email', email)]:
if not researcher.__dict__[kw] and val:
researcher.__dict__[kw] = val
save = True
if save:
researcher.save()
for variant in profile.other_names:
confidence = name_similarity(variant, variant)
name = Name.lookup_name(variant)
researcher.add_name_variant(name, confidence)
return researcher
def get_or_create_by_orcid(cls, orcid, profile=None, user=None):
"""
Creates (or returns an existing) researcher from its ORCID id.
:param profile: an :class:`OrcidProfile` object if it has already been fetched
from the API (otherwise we will fetch it ourselves)
:param user: an user to associate with the profile.
:returns: a :class:`Researcher` if everything went well, raises MetadataSourceException otherwise
"""
researcher = None
if orcid is None:
raise MetadataSourceException('Invalid ORCID id')
try:
researcher = Researcher.objects.get(orcid=orcid)
except Researcher.DoesNotExist:
if profile is None:
profile = OrcidProfile(id=orcid)
else:
profile = OrcidProfile(json=profile)
name = profile.name
homepage = profile.homepage
email = profile.email
researcher = Researcher.create_by_name(name[0],name[1], orcid=orcid,
user=user, homepage=homepage, email=email)
# Ensure that extra info is added.
save = False
for kw, val in [('homepage',homepage),('orcid',orcid),('email',email)]:
if not researcher.__dict__[kw] and val:
researcher.__dict__[kw] = val
save = True
if save:
researcher.save()
for variant in profile.other_names:
confidence = name_similarity(variant, variant)
name = Name.lookup_name(variant)
researcher.add_name_variant(name, confidence)
return researcher
def test_multiple(self):
self.assertAlmostEqual(
name_similarity(('Juan Pablo', 'Corella'),
('J. Pablo', 'Corella')), 1.0)
def test_multiple(self):
self.assertAlmostEqual(
name_similarity(('Juan Pablo', 'Corella'), ('J. Pablo', 'Corella')), 1.0)
def test_reverse(self):
self.assertAlmostEqual(
name_similarity(('W. Timothy', 'Gowers'), ('Timothy', 'Gowers')), 0.7)
def test_reverse(self):
self.assertAlmostEqual(
name_similarity(('W. Timothy', 'Gowers'), ('Timothy', 'Gowers')),
0.7)
