BertDefaultStringTextTermsMapper -- non masked entity values might be with separation between words

[nicolay-r]

In a nutshell, there is a class StringWithEmbeddingNetworkTermMapping, in which we cover this as follows:

AREkit/arekit/contrib/networks/core/input/terms_mapping.py

Lines 45 to 56 in 4a5b638

	def map_entity(self, e_ind, entity):
	assert(isinstance(entity, Entity))
	# Value extraction
	str_formatted_entity = super(StringWithEmbeddingNetworkTermMapping, self).map_entity(
	e_ind=e_ind,
	entity=entity)
	# Vector extraction
	emb_word, vector = self.__vectorizers[TermTypes.ENTITY].create_term_embedding(term=str_formatted_entity)
	return emb_word, vector

where create_term_embedding function provides an opportunity to replace with -.

The same does not work for BERT.