def __init__(self, session, node, parent):
SimpleTokenMerger.__init__(self, session, node, parent)
if nltk is None:
raise MissingDependencyException(self.objectType, 'nltk')
types = self.get_setting(session, 'entityTypes')
if types:
self.types = []
for type_ in types.split():
type_ = type_.lower()
if type_.startswith('pe'):
self.types.append('PERSON')
elif type_.startswith(('pl', 'g')):
self.types.append('GPE')
elif type_.startswith(('org', 'co')):
self.types.append('ORGANIZATION')
else:
msg = ("Unknown entity type setting {0} on {1} {2}"
"".format(type_,
self.__class__.__name__,
self.id)
)
raise ConfigFileException(msg)
else:
# Default to all
self.types = ['PERSON', 'GPE', 'ORGANIZATION']
# Should we keep the /POS tag or strip it
self.keepPos = self.get_setting(session, 'pos', 0)
def __init__(self, session, config, parent):
SimpleTokenMerger.__init__(self, session, config, parent)
self.nonPhrases = self.get_setting(session, "nonPhrases", 0)
self.nounRequired = self.get_setting(session, "nounRequired", 0)
def __init__(self, session, config, parent):
SimpleTokenMerger.__init__(self, session, config, parent)
self.nonPhrases = self.get_setting(session, 'nonPhrases', 0)
self.nounRequired = self.get_setting(session, 'nounRequired', 0)