def word2vec_embedding_from_sentences_v2(sentences,
CONFIGURATION,
sg=0,
size=100,
window=100):
from cle.wordembedding import EmbeddingHelper
model = EmbeddingHelper.embed(sentences, size, CONFIGURATION, window=100)
return model.wv
def train(documents, DIMENSIONS):
if len(documents) < 1:
return None
from cle.wordembedding import EmbeddingHelper
models = dict()
for predicate, literals in documents.items():
#literals = [TaggedDocument(doc, [i]) for i, doc in enumerate(literals)]
models[predicate] = EmbeddingHelper.embed(documents, DIMENSIONS, CONFIGURATION)
#models[predicate].build_vocab(literals)
#models[predicate].train(literals, total_examples=models[predicate].corpus_count, epochs=models[predicate].epochs)
#models[predicate].delete_temporary_training_data(keep_doctags_vectors=True, keep_inference=True)
return models
def train(documents, DIMENSIONS):
if len(documents) < 1:
return None
from cle.wordembedding import EmbeddingHelper
model = EmbeddingHelper.embed(documents, DIMENSIONS, CONFIGURATION)
import pandas as pd
# Reduce dimensionality
vecs = list()
#ids1 = np.random.choice(np.array(list(graph1.elements.keys())), size=min(10000,len(graph1.elements.keys())))
#ids2 = np.random.choice(np.array(list(graph2.elements.keys())), size=min(10000,len(graph2.elements.keys())))
#ids1 = np.array(list(graph1.elements.keys()))
#ids2 = np.array(list(graph2.elements.keys()))
ids1 = list()
ids2 = list()
return model
def train(documents, DIMENSIONS):
if len(documents) < 1:
return None
from cle.wordembedding import EmbeddingHelper
model = EmbeddingHelper.embed(documents, DIMENSIONS, CONFIGURATION)
return model