#335 fixed and #308 . Considering a function which calcuales the vect…

…or from parts rather than seeking such in an Embedding

arekit/contrib/networks/core/input/ctx_serialization.py

@@ -24,8 +24,4 @@ def WordEmbedding(self):
 
     @property
     def PosTagger(self):
-        raise NotImplementedError()
-
-    @property
-    def StringEntityEmbeddingFormatter(self):
         raise NotImplementedError()

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

@@ -2,8 +2,6 @@
 
 from arekit.common.data.input.terms_mapper import OpinionContainingTextTermsMapper
 from arekit.common.entities.base import Entity
-from arekit.common.entities.str_fmt import StringEntitiesFormatter
-from arekit.common.entities.types import OpinionEntityType
 from arekit.common.frames.text_variant import TextFrameVariant
 from arekit.contrib.networks.core.input.embedding.custom import create_term_embedding
 from arekit.contrib.networks.embeddings.base import Embedding
@@ -17,20 +15,18 @@ class StringWithEmbeddingNetworkTermMapping(OpinionContainingTextTermsMapper):
     MAX_PART_CUSTOM_EMBEDDING_SIZE = 3
     TOKEN_RANDOM_SEED_OFFSET = 12345
 
-    def __init__(self, predefined_embedding, string_entities_formatter, string_emb_entity_formatter):
+    def __init__(self, predefined_embedding, string_entities_formatter):
         """
         predefined_embedding:
         string_emb_entity_formatter:
             Utilized in order to obtain embedding value from predefined_embeding for enties
         """
         assert(isinstance(predefined_embedding, Embedding))
-        assert(isinstance(string_emb_entity_formatter, StringEntitiesFormatter))
 
         super(StringWithEmbeddingNetworkTermMapping, self).__init__(
             entity_formatter=string_entities_formatter)
 
         self.__predefined_embedding = predefined_embedding
-        self.__string_emb_entity_formatter = string_emb_entity_formatter
 
     def map_word(self, w_ind, word):
         value, vector = create_term_embedding(term=word,
@@ -67,23 +63,15 @@ def map_entity(self, e_ind, entity):
         assert(isinstance(entity, Entity))
 
         # Value extraction
-        str_entity_mask = super(StringWithEmbeddingNetworkTermMapping, self).map_entity(
+        str_formatted_entity = super(StringWithEmbeddingNetworkTermMapping, self).map_entity(
             e_ind=e_ind,
             entity=entity)
 
-        # TODO. Use synonyms. (Or use this functionality in a base class).
-        empty_set = set()
-        e_type = self.__get_entity_type(e_ind=e_ind,
-                                        subj_ind_set=empty_set,
-                                        obj_ind_set=empty_set)
-
         # Vector extraction
-        entity_word = self.__string_emb_entity_formatter.to_string(original_value=None,
-                                                                   entity_type=e_type)
-        m_ind = self.__predefined_embedding.try_find_index_by_plain_word(entity_word)
-        vector = self.__predefined_embedding.get_vector_by_index(m_ind)
+        emb_word, vector = create_term_embedding(term=str_formatted_entity,
+                                                 embedding=self.__predefined_embedding)
 
-        return str_entity_mask, vector
+        return emb_word, vector
 
     # region private methods
 
@@ -94,18 +82,4 @@ def __get_random_normal_distribution(vector_size, seed, loc, scale):
         np.random.seed(seed)
         return np.random.normal(loc=loc, scale=scale, size=vector_size)
 
-    @staticmethod
-    def __get_entity_type(e_ind, subj_ind_set, obj_ind_set):
-        assert(isinstance(e_ind, int))
-        assert(isinstance(subj_ind_set, set))
-        assert(isinstance(obj_ind_set, set))
-
-        result = OpinionEntityType.Other
-        if e_ind in obj_ind_set:
-            result = OpinionEntityType.Object
-        elif e_ind in subj_ind_set:
-            result = OpinionEntityType.Subject
-
-        return result
-
     # endregion

arekit/contrib/networks/handlers/serializer.py

@@ -87,8 +87,7 @@ def on_iteration(self, iter_index):
 
         text_terms_mapper = StringWithEmbeddingNetworkTermMapping(
             predefined_embedding=self.__exp_ctx.WordEmbedding,
-            string_entities_formatter=self.__exp_ctx.StringEntityFormatter,
-            string_emb_entity_formatter=self.__exp_ctx.StringEntityEmbeddingFormatter)
+            string_entities_formatter=self.__exp_ctx.StringEntityFormatter)
 
         text_provider = NetworkSingleTextProvider(
             text_terms_mapper=text_terms_mapper,