def names_are_substrings(name1, name2):
'''
Checks if two names are substrings of each other; e.g. "Christoph" vs. "Ch"
Only checks for the beginning of the names.
@param name1: Name string of the first name (w/ last name)
@type name1: string
@param name2: Name string of the second name (w/ last name)
@type name2: string
@return: are names synonymous
@rtype: boolean
'''
if not isinstance(name1, list):
name1 = split_name_parts(name1)
if not isinstance(name2, list):
name2 = split_name_parts(name2)
onames = name1[2]
tnames = name2[2]
# oname = "".join(onames).lower()
# tname = "".join(tnames).lower()
oname = clean_name_string("".join(onames).lower(), "", False, True)
tname = clean_name_string("".join(tnames).lower(), "", False, True)
names_are_substrings_b = False
if (oname.startswith(tname) or tname.startswith(oname)):
names_are_substrings_b = True
return names_are_substrings_b
def names_are_equal_composites(name1, name2):
'''
Checks if names are equal composites; e.g. "guangsheng" vs. "guang sheng"
@param name1: Name string of the first name (w/ last name)
@type name1: string
@param name2: Name string of the second name (w/ last name)
@type name2: string
@return: Are the names equal composites?
@rtype: boolean
'''
if not isinstance(name1, list):
name1 = split_name_parts(name1)
if not isinstance(name2, list):
name2 = split_name_parts(name2)
is_equal_composite = False
oname_variations = create_name_tuples(name1[2])
tname_variations = create_name_tuples(name2[2])
for oname_variation in oname_variations:
for tname_variation in tname_variations:
oname = clean_name_string(oname_variation.lower(), "", False, True)
tname = clean_name_string(tname_variation.lower(), "", False, True)
if oname == tname:
is_equal_composite = True
break
return is_equal_composite
def names_are_substrings(name1, name2):
'''
Checks if two names are substrings of each other; e.g. "Christoph" vs. "Ch"
Only checks for the beginning of the names.
@param name1: Name string of the first name (w/ last name)
@type name1: string
@param name2: Name string of the second name (w/ last name)
@type name2: string
@return: are names synonymous
@rtype: boolean
'''
if not isinstance(name1, list):
name1 = split_name_parts(name1)
if not isinstance(name2, list):
name2 = split_name_parts(name2)
onames = name1[2]
tnames = name2[2]
# oname = "".join(onames).lower()
# tname = "".join(tnames).lower()
oname = clean_name_string("".join(onames).lower(), "", False, True)
tname = clean_name_string("".join(tnames).lower(), "", False, True)
names_are_substrings_b = False
if (oname.startswith(tname)
or tname.startswith(oname)):
names_are_substrings_b = True
return names_are_substrings_b
def soft_compare_names(origin_name, target_name):
'''
Soft comparison of names, to use in search engine an similar
Base results:
If surname is equal in [0.6,1.0]
If surname similar in [0.4,0.8]
If surname differs in [0.0,0.4]
all depending on average compatibility of names and initials.
'''
jaro_fctn = None
try:
from Levenshtein import jaro_winkler
jaro_fctn = jaro_winkler
except ImportError:
jaro_fctn = jaro_winkler_str_similarity
score = 0.0
oname = deepcopy(origin_name)
tname = deepcopy(target_name)
orig_name = split_name_parts(oname.lower())
targ_name = split_name_parts(tname.lower())
orig_name[0] = clean_name_string(orig_name[0],
replacement="",
keep_whitespace=False)
targ_name[0] = clean_name_string(targ_name[0],
replacement="",
keep_whitespace=False)
if orig_name[0] == targ_name[0]:
score += 0.6
else:
if ((jaro_fctn(orig_name[0].lower(), targ_name[0].lower()) < .95)
or min(len(orig_name[0]), len(targ_name[0])) <= 4):
score += 0.0
else:
score += 0.4
if orig_name[1] and targ_name[1]:
max_initials = max(len(orig_name[1]), len(targ_name[1]))
matching_i = 0
if len(orig_name[1]) >= 1 and len(targ_name[1]) >= 1:
for i in orig_name[1]:
if i in targ_name[1]:
matching_i += 1
max_names = max(len(orig_name[2]), len(targ_name[2]))
matching_n = 0
if len(orig_name[2]) >= 1 and len(targ_name[2]) >= 1:
cleaned_targ_name = [
clean_name_string(i, replacement="", keep_whitespace=False)
for i in targ_name[2]
]
for i in orig_name[2]:
if clean_name_string(
i, replacement="",
keep_whitespace=False) in cleaned_targ_name:
matching_n += 1
name_score = (matching_i + matching_n) * 0.4 / (max_names +
max_initials)
score += name_score
return score
def names_are_equal_composites(name1, name2):
'''
Checks if names are equal composites; e.g. "guangsheng" vs. "guang sheng"
@param name1: Name string of the first name (w/ last name)
@type name1: string
@param name2: Name string of the second name (w/ last name)
@type name2: string
@return: Are the names equal composites?
@rtype: boolean
'''
if not isinstance(name1, list):
name1 = split_name_parts(name1)
if not isinstance(name2, list):
name2 = split_name_parts(name2)
is_equal_composite = False
oname_variations = create_name_tuples(name1[2])
tname_variations = create_name_tuples(name2[2])
for oname_variation in oname_variations:
for tname_variation in tname_variations:
oname = clean_name_string(oname_variation.lower(), "", False, True)
tname = clean_name_string(tname_variation.lower(), "", False, True)
if oname == tname:
is_equal_composite = True
break
return is_equal_composite
def test_clean_name_string(self):
"""bibauthorid - test cleaning of name strings"""
self.assertEqual('this is a full name',
baidu.clean_name_string('this is a full name'))
self.assertEqual('this is a full ,. pz',
baidu.clean_name_string('this is a full ;,.$&[{{}}(=*)+]pz'))
self.assertEqual('',
baidu.clean_name_string(''))
def soft_compare_names(origin_name, target_name):
'''
Soft comparison of names, to use in search engine an similar
Base results:
If surname is equal in [0.6,1.0]
If surname similar in [0.4,0.8]
If surname differs in [0.0,0.4]
all depending on average compatibility of names and initials.
'''
jaro_fctn = None
try:
from Levenshtein import jaro_winkler
jaro_fctn = jaro_winkler
except ImportError:
jaro_fctn = jaro_winkler_str_similarity
score = 0.0
oname = deepcopy(origin_name)
tname = deepcopy(target_name)
orig_name = split_name_parts(oname.lower())
targ_name = split_name_parts(tname.lower())
orig_name[0] = clean_name_string(orig_name[0],
replacement="",
keep_whitespace=False)
targ_name[0] = clean_name_string(targ_name[0],
replacement="",
keep_whitespace=False)
if orig_name[0] == targ_name[0]:
score += 0.6
else:
if ((jaro_fctn(orig_name[0].lower(), targ_name[0].lower()) < .95)
or min(len(orig_name[0]), len(targ_name[0])) <= 4):
score += 0.0
else:
score += 0.4
if orig_name[1] and targ_name[1]:
max_initials = max(len(orig_name[1]), len(targ_name[1]))
matching_i = 0
if len(orig_name[1]) >= 1 and len(targ_name[1]) >= 1:
for i in orig_name[1]:
if i in targ_name[1]:
matching_i += 1
max_names = max(len(orig_name[2]), len(targ_name[2]))
matching_n = 0
if len(orig_name[2]) >= 1 and len(targ_name[2]) >= 1:
cleaned_targ_name = [clean_name_string(i, replacement="", keep_whitespace=False) for i in targ_name[2]]
for i in orig_name[2]:
if clean_name_string(i, replacement="", keep_whitespace=False) in cleaned_targ_name:
matching_n += 1
name_score = (matching_i + matching_n) * 0.4 / (max_names + max_initials)
score += name_score
return score
def names_are_equal_gender(name1, name2, gendernames):
'''
Checks on gender equality of two names baes on a word list
@param name1: Name string of the first name (w/ last name)
@type name1: string
@param name2: Name string of the second name (w/ last name)
@type name2: string
@param gendernames: dictionary of male/female names
@type gendernames: dict
@return: Are names gender-equal?
@rtype: boolean
'''
if not isinstance(name1, list):
name1 = split_name_parts(name1)
if not isinstance(name2, list):
name2 = split_name_parts(name2)
print_debug = False
names_are_equal_gender_b = True
ogender = None
tgender = None
oname = name1[2][0].lower()
tname = name2[2][0].lower()
oname = clean_name_string(oname, "", False, True)
tname = clean_name_string(tname, "", False, True)
if oname in gendernames['boys']:
ogender = 'Male'
elif oname in gendernames['girls']:
ogender = 'Female'
if tname in gendernames['boys']:
tgender = 'Male'
elif tname in gendernames['girls']:
tgender = 'Female'
if print_debug:
print ' Gender check: ', oname, ' is a ', ogender
print ' Gender check: ', tname, ' is a ', tgender
if ogender and tgender:
if ogender != tgender:
if print_debug:
print ' Gender differs, force split!'
names_are_equal_gender_b = False
return names_are_equal_gender_b
def names_are_equal_gender(name1, name2, gendernames):
'''
Checks on gender equality of two names baes on a word list
@param name1: Name string of the first name (w/ last name)
@type name1: string
@param name2: Name string of the second name (w/ last name)
@type name2: string
@param gendernames: dictionary of male/female names
@type gendernames: dict
@return: Are names gender-equal?
@rtype: boolean
'''
if not isinstance(name1, list):
name1 = split_name_parts(name1)
if not isinstance(name2, list):
name2 = split_name_parts(name2)
print_debug = False
names_are_equal_gender_b = True
ogender = None
tgender = None
oname = name1[2][0].lower()
tname = name2[2][0].lower()
oname = clean_name_string(oname, "", False, True)
tname = clean_name_string(tname, "", False, True)
if oname in gendernames['boys']:
ogender = 'Male'
elif oname in gendernames['girls']:
ogender = 'Female'
if tname in gendernames['boys']:
tgender = 'Male'
elif tname in gendernames['girls']:
tgender = 'Female'
if print_debug:
print ' Gender check: ', oname, ' is a ', ogender
print ' Gender check: ', tname, ' is a ', tgender
if ogender and tgender:
if ogender != tgender:
if print_debug:
print ' Gender differs, force split!'
names_are_equal_gender_b = False
return names_are_equal_gender_b
def update_virtualauthor_cluster(va_id):
"""
Computes the clusterID for the virtualauthor.
The clustering at this level is done accumulating all the compatible
surnames, but there is
still space for a smarter choice.
@note: The clustering number is used as a speed up tool, and it is meant to
accumulate all the possible compatible virtualauthors so we do not have
to search anywhere else. This means that a worst case cluster is no
clustering at all (only one cluster for the whole virtualauthors set)
and a best case cluster is a clusterID for the minimum set of
virtualauthors representing a real author. It is important
_NOT to loose_ virtualauthors! It is better to have a bigger cluster
than loosing possibly precious information later.
@param va_id: ID of the virtual author the cluster shall be updated for
@type va_id: int
"""
ori_name = get_virtualauthor_records(va_id, 'orig_name_string')[0]['value']
ori_name = clean_name_string(ori_name)
current_cluster_ids = get_clusterids_from_name(ori_name, True)
for va_item in [row for row in dat.VIRTUALAUTHORS
if row['virtualauthorid'] == va_id]:
va_item['clusterid'] = current_cluster_ids[1]
bconfig.LOGGER.debug("| Found %s cluster for %s. Now set to %s." %
(len(current_cluster_ids[0]), ori_name, current_cluster_ids[1]))
update_virtualauthor_record(va_id, 'updated', 'True')
def update_virtualauthor_cluster(va_id):
"""
Computes the clusterID for the virtualauthor.
The clustering at this level is done accumulating all the compatible
surnames, but there is
still space for a smarter choice.
@note: The clustering number is used as a speed up tool, and it is meant to
accumulate all the possible compatible virtualauthors so we do not have
to search anywhere else. This means that a worst case cluster is no
clustering at all (only one cluster for the whole virtualauthors set)
and a best case cluster is a clusterID for the minimum set of
virtualauthors representing a real author. It is important
_NOT to loose_ virtualauthors! It is better to have a bigger cluster
than loosing possibly precious information later.
@param va_id: ID of the virtual author the cluster shall be updated for
@type va_id: int
"""
ori_name = get_virtualauthor_records(va_id, 'orig_name_string')[0]['value']
ori_name = clean_name_string(ori_name)
current_cluster_ids = get_clusterids_from_name(ori_name, True)
for va_item in [
row for row in dat.VIRTUALAUTHORS
if row['virtualauthorid'] == va_id
]:
va_item['clusterid'] = current_cluster_ids[1]
bconfig.LOGGER.debug(
"| Found %s cluster for %s. Now set to %s." %
(len(current_cluster_ids[0]), ori_name, current_cluster_ids[1]))
update_virtualauthor_record(va_id, 'updated', 'True')
def names_are_synonymous(name1, name2, name_variations):
'''
Checks if two names are synonymous; e.g. "Robert" vs. "Bob"
@param name1: Name string of the first name (w/ last name)
@type name1: string
@param name2: Name string of the second name (w/ last name)
@type name2: string
@param name_variations: name variations list
@type name_variations: list of lists
@return: are names synonymous
@rtype: boolean
'''
if not isinstance(name1, list):
name1 = split_name_parts(name1)
if not isinstance(name2, list):
name2 = split_name_parts(name2)
print_debug = False
names_are_synonymous_b = False
max_matches = min(len(name1[2]), len(name2[2]))
matches = []
for i in xrange(max_matches):
matches.append(False)
for nvar in name_variations:
for i in xrange(max_matches):
oname = name1[2][i].lower()
tname = name2[2][i].lower()
oname = clean_name_string(oname, "", False, True)
tname = clean_name_string(tname, "", False, True)
if oname in nvar and tname in nvar:
if print_debug:
print ' ', oname, ' and ', tname, ' are synonyms! Not splitting!'
matches[i] = True
if sum(matches) == max_matches:
names_are_synonymous_b = True
break
return names_are_synonymous_b
def names_are_synonymous(name1, name2, name_variations):
'''
Checks if two names are synonymous; e.g. "Robert" vs. "Bob"
@param name1: Name string of the first name (w/ last name)
@type name1: string
@param name2: Name string of the second name (w/ last name)
@type name2: string
@param name_variations: name variations list
@type name_variations: list of lists
@return: are names synonymous
@rtype: boolean
'''
if not isinstance(name1, list):
name1 = split_name_parts(name1)
if not isinstance(name2, list):
name2 = split_name_parts(name2)
print_debug = False
names_are_synonymous_b = False
max_matches = min(len(name1[2]), len(name2[2]))
matches = []
for i in xrange(max_matches):
matches.append(False)
for nvar in name_variations:
for i in xrange(max_matches):
oname = name1[2][i].lower()
tname = name2[2][i].lower()
oname = clean_name_string(oname, "", False, True)
tname = clean_name_string(tname, "", False, True)
if oname in nvar and tname in nvar:
if print_debug:
print ' ', oname, ' and ', tname, ' are synonyms! Not splitting!'
matches[i] = True
if sum(matches) == max_matches:
names_are_synonymous_b = True
break
return names_are_synonymous_b
def names_minimum_levenshtein_distance(name1, name2):
'''
Determines the minimum distance D between two names.
Comparison is base on the minimum number of first names.
Examples:
D("guang", "guang sheng") = 0
D("guang", "guangsheng") = 5
D("guang sheng", "guangsheng") = 5
D("guang sheng", "guang shing") = 1
D("guang ming", "guang fin") = 2
@precondition: Names have been checked for composition equality.
@param name1: Name string of the first name (w/ last name)
@type name1: string
@param name2: Name string of the second name (w/ last name)
@type name2: string
@return: the minimum Levenshtein distance between two names
@rtype: int
'''
try:
from Levenshtein import distance
except ImportError:
bconfig.LOGGER.exception("Levenshtein Module not available!")
return - 1
if not isinstance(name1, list):
name1 = split_name_parts(name1)
if not isinstance(name2, list):
name2 = split_name_parts(name2)
onames = name1[2]
tnames = name2[2]
# min_names_count = min(len(onames), len(tnames))
#
# if min_names_count <= 0:
# return -1
#
# oname = "".join(onames[:min_names_count]).lower()
# tname = "".join(tnames[:min_names_count]).lower()
oname = clean_name_string("".join(onames).lower(), "", False, True)
tname = clean_name_string("".join(tnames).lower(), "", False, True)
return distance(oname, tname)
def names_minimum_levenshtein_distance(name1, name2):
'''
Determines the minimum distance D between two names.
Comparison is base on the minimum number of first names.
Examples:
D("guang", "guang sheng") = 0
D("guang", "guangsheng") = 5
D("guang sheng", "guangsheng") = 5
D("guang sheng", "guang shing") = 1
D("guang ming", "guang fin") = 2
@precondition: Names have been checked for composition equality.
@param name1: Name string of the first name (w/ last name)
@type name1: string
@param name2: Name string of the second name (w/ last name)
@type name2: string
@return: the minimum Levenshtein distance between two names
@rtype: int
'''
try:
from Levenshtein import distance
except ImportError:
bconfig.LOGGER.exception("Levenshtein Module not available!")
return -1
if not isinstance(name1, list):
name1 = split_name_parts(name1)
if not isinstance(name2, list):
name2 = split_name_parts(name2)
onames = name1[2]
tnames = name2[2]
# min_names_count = min(len(onames), len(tnames))
#
# if min_names_count <= 0:
# return -1
#
# oname = "".join(onames[:min_names_count]).lower()
# tname = "".join(tnames[:min_names_count]).lower()
oname = clean_name_string("".join(onames).lower(), "", False, True)
tname = clean_name_string("".join(tnames).lower(), "", False, True)
return distance(oname, tname)
def get_clusterids_from_name(name, return_matching=False):
'''
Returns a list of cluster IDs, which are fitting for the parameter 'name'.
First checks if, in general, a cluster for this name exists. If not,
create one. If there is a cluster, try to find all other fitting clusters
and add the found cluster IDs to the list to be returned
@param name: The name to be on the lookout for.
@type name: string
@param return_matching: also return the reference name's matching cluster
@type return_matching: boolean
@return:
if return_matching: list of 1) list of cluster IDs 2) the cluster ID
matching the name
if not return_matching: list of cluster IDs
@rtype:
if return_matching: list of (list of int, int)
if not return_matching: list of int
'''
search_string = create_unified_name(name)
search_string = clean_name_string(search_string)
if len(search_string) > 150:
search_string = search_string[:150]
clusterids = set()
matching_cluster = -1
initials = ""
split_string = ""
if search_string[:-1].count(",") > 0:
split_string = search_string[:-1].replace(' ', '').split(',')
if split_string[1]:
initials = split_string[1].split('.')
if len(initials) > 2 and len(initials) <= 5:
permutation_list = initials
permutation_base = ("%s, %s." %
(search_string.split(',')[0], permutation_list[0]))
for permutation in permutations(permutation_list[1:]):
name_string = "%s %s." % (permutation_base, ". ".join(permutation))
clusters = _get_clusterids_from_name(name_string, return_matching)
if return_matching:
matching_cluster = clusters[1]
for clusterid in clusters[0]:
clusterids.add(clusterid)
else:
for clusterid in clusters:
clusterids.add(clusterid)
else:
clusters = _get_clusterids_from_name(search_string, return_matching)
if return_matching:
matching_cluster = clusters[1]
clusterids = clusters[0]
else:
clusterids = clusters
if return_matching:
return [clusterids, matching_cluster]
else:
return clusterids
def compare_names(origin_name, target_name):
"""
Compute an index of confidence that would like to indicate whether two
names might represent the same person.The computation is based on
similarities of name structure, in particular:
Initials:
We assign an high score if all the initials matches are in the
right order, much lower if they are in the wrong order
Names:
We assign a lower score for mismatching names and higher score for
fully matching names
If there is nothing to compare we are forced to assume a high score.
Example for splitting names:
In : bibauthorid.split_name_parts("Ellis, John R")
Out: ['Ellis', ['J', 'R'], ['John']]
Ellis, R. Keith => [ [Ellis], [R, K], [Keith] ]
Ellis, Richard Keith => [ [Ellis], [R, K], [Richard, Keith] ]
Since the initials are computed whether on the real initials present in the
name string and using the full name, if there is no initials match we are 1
00% confident that:
1. we have no names/initials at all, or
2. we have completely different names; hence if there is no initial
match we skip this step.
@param orig_name: The first author's last name, first name(s) and initial
@type orig_name: list of strings and lists of strings
@param targ_name: The second author's last name, first name(s) and initial
@type targ_name: list of strings and lists of strings
@return: a value that describes the likelihood of the names being the same
@rtype: float
"""
jaro_fctn = None
try:
from Levenshtein import jaro_winkler
jaro_fctn = jaro_winkler
except ImportError:
jaro_fctn = jaro_winkler_str_similarity
oname = deepcopy(origin_name)
tname = deepcopy(target_name)
orig_name = split_name_parts(oname.lower())
targ_name = split_name_parts(tname.lower())
bconfig.LOGGER.info("|--> Comparing Names: \"%s\" and \"%s\"" %
(origin_name, target_name))
lastname_modifier = 0.0
if not (orig_name[0] == targ_name[0]):
# last names are not equal before cleaning them. Assign entry penalty.
lastname_modifier = 0.15
orig_name[0] = clean_name_string(orig_name[0],
replacement="",
keep_whitespace=False)
targ_name[0] = clean_name_string(targ_name[0],
replacement="",
keep_whitespace=False)
if not (orig_name[0] == targ_name[0]):
if ((jaro_fctn(orig_name[0].lower(), targ_name[0].lower()) < .95)
or min(len(orig_name[0]), len(targ_name[0])) <= 4):
bconfig.LOGGER.warn(("Unequal lastnames(%s vs. %s)."
+ "Skipping Comparison")
% (orig_name[0], targ_name[0]))
return 0.0
else:
bconfig.LOGGER.log(25, "Last names are not equal; "
+ "but similar enough to continue the comparison")
# Let it go through...however, reduce the final result a little.
lastname_modifier = 0.24
else:
# last names are equal after cleaning them. Reduce penalty.
if lastname_modifier == 0.15:
lastname_modifier = 0.02
if orig_name[2] and targ_name[2]:
if len(orig_name[2]) > 1 or len(targ_name[2]) > 1:
variation_ps = []
oname_variations = create_name_tuples(orig_name[2])
tname_variations = create_name_tuples(targ_name[2])
for oname_variation in oname_variations:
for tname_variation in tname_variations:
oname_var = split_name_parts("%s, %s"
% (orig_name[0],
oname_variation))
tname_var = split_name_parts("%s, %s"
% (targ_name[0],
tname_variation))
variation_ps.append(_perform_matching(oname_var,
tname_var))
return max(variation_ps) - lastname_modifier
return _perform_matching(orig_name, targ_name) - lastname_modifier
def get_clusterids_from_name(name, return_matching=False):
'''
Returns a list of cluster IDs, which are fitting for the parameter 'name'.
First checks if, in general, a cluster for this name exists. If not,
create one. If there is a cluster, try to find all other fitting clusters
and add the found cluster IDs to the list to be returned
@param name: The name to be on the lookout for.
@type name: string
@param return_matching: also return the reference name's matching cluster
@type return_matching: boolean
@return:
if return_matching: list of 1) list of cluster IDs 2) the cluster ID
matching the name
if not return_matching: list of cluster IDs
@rtype:
if return_matching: list of (list of int, int)
if not return_matching: list of int
'''
search_string = create_unified_name(name)
search_string = clean_name_string(search_string)
if len(search_string) > 150:
search_string = search_string[:150]
clusterids = set()
matching_cluster = -1
initials = ""
split_string = ""
if search_string[:-1].count(",") > 0:
split_string = search_string[:-1].replace(' ', '').split(',')
if split_string[1]:
initials = split_string[1].split('.')
if len(initials) > 2 and len(initials) <= 5:
permutation_list = initials
permutation_base = ("%s, %s." %
(search_string.split(',')[0], permutation_list[0]))
for permutation in permutations(permutation_list[1:]):
name_string = "%s %s." % (permutation_base, ". ".join(permutation))
clusters = _get_clusterids_from_name(name_string, return_matching)
if return_matching:
matching_cluster = clusters[1]
for clusterid in clusters[0]:
clusterids.add(clusterid)
else:
for clusterid in clusters:
clusterids.add(clusterid)
else:
clusters = _get_clusterids_from_name(search_string, return_matching)
if return_matching:
matching_cluster = clusters[1]
clusterids = clusters[0]
else:
clusterids = clusters
if return_matching:
return [clusterids, matching_cluster]
else:
return clusterids
def compare_names(origin_name, target_name):
"""
Compute an index of confidence that would like to indicate whether two
names might represent the same person.The computation is based on
similarities of name structure, in particular:
Initials:
We assign an high score if all the initials matches are in the
right order, much lower if they are in the wrong order
Names:
We assign a lower score for mismatching names and higher score for
fully matching names
If there is nothing to compare we are forced to assume a high score.
Example for splitting names:
In : bibauthorid.split_name_parts("Ellis, John R")
Out: ['Ellis', ['J', 'R'], ['John']]
Ellis, R. Keith => [ [Ellis], [R, K], [Keith] ]
Ellis, Richard Keith => [ [Ellis], [R, K], [Richard, Keith] ]
Since the initials are computed whether on the real initials present in the
name string and using the full name, if there is no initials match we are 1
00% confident that:
1. we have no names/initials at all, or
2. we have completely different names; hence if there is no initial
match we skip this step.
@param orig_name: The first author's last name, first name(s) and initial
@type orig_name: list of strings and lists of strings
@param targ_name: The second author's last name, first name(s) and initial
@type targ_name: list of strings and lists of strings
@return: a value that describes the likelihood of the names being the same
@rtype: float
"""
jaro_fctn = None
try:
from Levenshtein import jaro_winkler
jaro_fctn = jaro_winkler
except ImportError:
jaro_fctn = jaro_winkler_str_similarity
oname = deepcopy(origin_name)
tname = deepcopy(target_name)
orig_name = split_name_parts(oname.lower())
targ_name = split_name_parts(tname.lower())
bconfig.LOGGER.info("|--> Comparing Names: \"%s\" and \"%s\"" %
(origin_name, target_name))
lastname_modifier = 0.0
if not (orig_name[0] == targ_name[0]):
# last names are not equal before cleaning them. Assign entry penalty.
lastname_modifier = 0.15
orig_name[0] = clean_name_string(orig_name[0],
replacement="",
keep_whitespace=False)
targ_name[0] = clean_name_string(targ_name[0],
replacement="",
keep_whitespace=False)
if not (orig_name[0] == targ_name[0]):
if ((jaro_fctn(orig_name[0].lower(), targ_name[0].lower()) < .95)
or min(len(orig_name[0]), len(targ_name[0])) <= 4):
bconfig.LOGGER.warn(
("Unequal lastnames(%s vs. %s)." + "Skipping Comparison") %
(orig_name[0], targ_name[0]))
return 0.0
else:
bconfig.LOGGER.log(
25, "Last names are not equal; " +
"but similar enough to continue the comparison")
# Let it go through...however, reduce the final result a little.
lastname_modifier = 0.24
else:
# last names are equal after cleaning them. Reduce penalty.
if lastname_modifier == 0.15:
lastname_modifier = 0.02
if orig_name[2] and targ_name[2]:
if len(orig_name[2]) > 1 or len(targ_name[2]) > 1:
variation_ps = []
oname_variations = create_name_tuples(orig_name[2])
tname_variations = create_name_tuples(targ_name[2])
for oname_variation in oname_variations:
for tname_variation in tname_variations:
oname_var = split_name_parts(
"%s, %s" % (orig_name[0], oname_variation))
tname_var = split_name_parts(
"%s, %s" % (targ_name[0], tname_variation))
variation_ps.append(_perform_matching(
oname_var, tname_var))
return max(variation_ps) - lastname_modifier
return _perform_matching(orig_name, targ_name) - lastname_modifier