短文字轉向量的一種實現方式
阿新 • • 發佈:2018-12-12
文章目錄
前言
下文實現僅僅是比較粗糙的一種方式,可以改進的點還有很多,是真的很多!重點是,不講解原理,就是這麼沒道理…
實現思路
- 分詞。分詞還是jieba好。word2vec模型訓練選取gensim。
- 使用大語料進行基礎詞典word2vec模型的訓練。
- 使用特定領域(針對業務)語料進行專業詞彙word2vec模型的訓練。
- 文字分詞後使用AVG-W2V方式獲取短文字向量,維度取決於word2vec維度大小,即所有詞向量求平均。
word2vec相關配置
- w2v.properties
#一些經驗 #架構(sg):skip-gram(慢、對罕見字有利)vs CBOW(快) #訓練演算法(hs):分層softmax(對罕見字有利)vs 負取樣(對常見詞和低緯向量有利) #欠取樣頻繁詞(sample):可以提高結果的準確性和速度(適用範圍1e-3到1e-5) #文字大小(window):skip-gram通常在10附近,CBOW通常在5附近 #大語料下,建議提高min_count,減少iter # 訓練演算法,0為CBOW演算法,1為skip-gram演算法,預設為0 sg=1 # 特徵向量的維度 size=300 # 詞窗大小 window=5 # 最小詞頻 min_count=5 # 初始學習速率 alpha=0.025 # 0為負取樣,1為softmax,預設為0 hs=1 #迭代次數 iter=10
程式碼
- 大語料基礎訓練相關程式碼
# -*- coding:utf-8 -*-
"""
Description: 基於百度百科大語料的word2vec模型
@author: WangLeAi
@date: 2018/9/18
"""
import os
from util.DBUtil import DbPoolUtil
from util.JiebaUtil import jieba_util
from util.PropertiesUtil import prop
from gensim.models import word2vec
class OriginModel(object) :
def __init__(self):
self.params = prop.get_config_dict("config/w2v.properties")
self.db_pool_util = DbPoolUtil(db_type="mysql")
self.train_data_path = "gen/ori_train_data.txt"
self.model_path = "model/oriw2v.model"
@staticmethod
def text_process(sentence):
"""
文字預處理
:param sentence:
:return:
"""
# 過濾任意非中文、非英文、非數字
# regex = re.compile(u'[^\u4e00-\u9fa50-9a-zA-Z\-·]+')
# sentence = regex.sub('', sentence)
words = jieba_util.jieba_cut(sentence)
return words
def get_train_data(self):
"""
獲取訓練資料,此處需要自行修改,最好寫入檔案而不是直接取到記憶體中!!!!!
:return:
"""
print("建立初始語料訓練資料")
sql = """ """
sentences = self.db_pool_util.loop_row(origin_model, "text_process", sql)
with open(self.train_data_path, "w", encoding="utf-8") as f:
for sentence in sentences:
f.write(" ".join(sentence) + "\n")
def train_model(self):
"""
訓練模型
:return:
"""
if not os.path.exists(self.train_data_path):
self.get_train_data()
print("訓練初始模型")
sentences = word2vec.LineSentence(self.train_data_path)
model = word2vec.Word2Vec(sentences=sentences, sg=int(self.params["sg"]), size=int(self.params["size"]),
window=int(self.params["window"]), min_count=int(self.params["min_count"]),
alpha=float(self.params["alpha"]), hs=int(self.params["hs"]), workers=6,
iter=int(self.params["iter"]))
model.save(self.model_path)
print("訓練初始模型完畢,儲存模型")
origin_model = OriginModel()
- 額外語料進行訓練
# -*- coding:utf-8 -*-
"""
Description:word2vec fine tuning
基於對應型別的額外語料進行微調
@author: WangLeAi
@date: 2018/9/11
"""
import os
from util.DBUtil import DbPoolUtil
from util.JiebaUtil import jieba_util
from util.PropertiesUtil import prop
from gensim.models import word2vec
from algorithms.OriginModel import origin_model
class Word2VecModel(object):
def __init__(self):
self.db_pool_util = DbPoolUtil(db_type="mysql")
self.train_data_path = "gen/train_data.txt"
self.origin_model_path = "model/oriw2v.model"
self.model_path = "model/w2v.model"
self.model = None
# 未登入詞進入需考慮最小詞頻
self.min_count = int(prop.get_config_value("config/w2v.properties", "min_count"))
@staticmethod
def text_process(sentence):
"""
文字預處理
:param sentence:
:return:
"""
# 過濾任意非中文、非英文、非數字等
# regex = re.compile(u'[^\u4e00-\u9fa50-9a-zA-Z\-·]+')
# sentence = regex.sub('', sentence)
words = jieba_util.jieba_cut(sentence)
return words
def get_train_data(self):
"""
獲取訓練資料,此處需要自行修改,最好寫入檔案而不是直接取到記憶體中!!!!!
:return:
"""
print("建立額外語料訓練資料")
sql = """ """
sentences = self.db_pool_util.loop_row(w2v_model, "text_process", sql)
with open(self.train_data_path, "a", encoding="utf-8") as f:
for sentence in sentences:
f.write(" ".join(sentence) + "\n")
def train_model(self):
"""
訓練模型
:return:
"""
if not os.path.exists(self.origin_model_path):
print("無初始模型,進行初始模型訓練")
origin_model.train_model()
model = word2vec.Word2Vec.load(self.origin_model_path)
print("初始模型載入完畢")
if not os.path.exists(self.train_data_path):
self.get_train_data()
print("額外語料訓練")
extra_sentences = word2vec.LineSentence(self.train_data_path)
model.build_vocab(extra_sentences, update=True)
model.train(extra_sentences, total_examples=model.corpus_count, epochs=model.iter)
model.save(self.model_path)
print("額外語料訓練完畢")
def load_model(self):
"""
載入模型
:return:
"""
print("載入詞嵌入模型")
if not os.path.exists(self.model_path):
print("無詞嵌入模型,進行訓練")
self.train_model()
self.model = word2vec.Word2Vec.load(self.model_path)
print("詞嵌入模型載入完畢")
def get_word_vector(self, words, extra=0):
"""
獲取詞語向量,需要先載入模型
:param words:
:param extra:是否考慮未登入詞,0不考慮,1考慮
:return:
"""
if extra:
if words not in self.model:
more_sentences = [[words, ] for i in range(self.min_count)]
self.model.build_vocab(more_sentences, update=True)
self.model.train(more_sentences, total_examples=self.model.corpus_count, epochs=self.model.iter)
self.model.save(self.model_path)
rst = None
if words in self.model:
rst = self.model[words]
return rst
def get_sentence_vector(self, sentence, extra=0):
"""
獲取文字向量,需要先載入模型
:param sentence:
:param extra: 是否考慮未登入詞,0不考慮,1考慮
:return:
"""
words = jieba_util.jieba_cut_flag(sentence)
if not words:
words = jieba_util.jieba_cut(sentence)
if not words:
print("存在無法切出有效詞的句子:" + sentence)
# raise Exception("存在無法切出有效詞的句子:" + sentence)
if extra:
for item in words:
if item not in self.model:
more_sentences = [words for i in range(self.min_count)]
self.model.build_vocab(more_sentences, update=True)
self.model.train(more_sentences, total_examples=self.model.corpus_count, epochs=self.model.iter)
self.model.save(self.model_path)
break
return self.get_sentence_embedding(words)
def get_sentence_embedding(self, words):
"""
獲取短文字向量,僅推薦短文字使用
句中所有詞權重總和求平均獲取文字向量,不適用於長文字的原因在於受頻繁詞影響較大
長文字推薦使用gensim的doc2vec
:param words:
:return:
"""
count = 0
vector = None
for item in words:
if item in self.model:
count += 1
if vector is not None:
vector = vector + self.model[item]
else:
vector = self.model[item]
if vector is not None:
vector = vector / count
return vector
w2v_model = Word2VecModel()
- 測試方式
# -*- coding:utf-8 -*-
"""
Description:
@author: WangLeAi
@date: 2018/9/18
"""
import os
from algorithms.Word2VecModel import w2v_model
def main():
root_path = os.path.split(os.path.realpath(__file__))[0]
if not os.path.exists(root_path + "/model"):
os.mkdir(root_path + "/model")
w2v_model.load_model()
print(w2v_model.get_sentence_vector("不知不覺間我已經忘記了愛"))
if __name__ == "__main__":
main()