word2ve源码解析(python版gensim实现)

class BaseAny2VecModel()

class BaseWordEmbeddingsModel(BaseAny2VecModel)

def __init__():
    super(BaseWordEmbeddingsModel, self).__init__()
	self.build_vocab(sentences)
	self.train()
	 
def build_vocab():
     total_words, corpus_count = self.vocabulary.scan_vocab()
	 report_values = self.vocabulary.prepare_vocab()

def train():

def similarity():

def most_similar():

class Word2Vec(BaseWordEmbeddingsModel)

 def __init__():
     
	 #wv: 是类 ~gensim.models.keyedvectors.Word2VecKeyedVectors生产的对象，在word2vec是一个属性

为了在不同的训练算法（Word2Vec，Fastext，WordRank，VarEmbed）之间共享单词向量查询代码，gensim将单词向量的存储和查询分离为一个单独的类 KeyedVectors包含单词和对应向量的映射。可以通过它进行词向量的查询
self.wv = Word2VecKeyedVectors(size)
self.vocabulary = Word2VecVocab()
super(Word2Vec, self).init()

 def train():
     return super(Word2Vec, self).train()

 #保存模型
 def save():
 #加载模型
 def load():

class Word2VecVocab()
def init():
#该方法是从句子序列中构建单词表
def scan_vocab():

 def sort_vocab():
 
 def prepare_vocab():
 
 #使用存储的词汇单词及其词频创建一个二进制哈夫曼树
 def create_binary_tree():

#训练词向量的内部浅层神经网络
class Word2VecTrainables():
def init():
def prepare_weights():
def seeded_vector():
def reset_weights():
def update_weights():