-
Notifications
You must be signed in to change notification settings - Fork 12
/
doc2vec-sim.py
127 lines (111 loc) · 4.29 KB
/
doc2vec-sim.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
# !/usr/bin/env python
# -*- coding:utf-8 _*-
"""
@Author:yanqiang
@File: doc2vec-kmens.py
@Time: 2018/11/12 16:12
@Software: PyCharm
@Description:
"""
import gensim
import numpy as np
from gensim.models.doc2vec import Doc2Vec,LabeledSentence
from sklearn.cluster import KMeans
import os
import pandas as pd
import pickle
import jieba
import random
def load_articles():
"""
将文本分词
:return:
"""
def split_artilce(article):
return [word for word in article.split(' ') if word != '']
filename='data/36kr_articles.pkl'
if os.path.exists(filename):
with open(filename,'rb') as in_data:
articles=pickle.load(in_data)
articles = [split_artilce(article) for article in articles]
return articles
stop_words = [word.strip() for word in open('data/stop_words.txt', 'r',encoding='utf-8').readlines()]
print(stop_words)
data = pd.read_csv('data/36kr_articles.csv')
print("正在分词...")
# 分词，去除停用词
data['title'] = data['title'].apply(lambda x: " ".join([word for word in jieba.cut(x)
if word not in stop_words and x]))
data['summary'] = data['summary'].apply(lambda x: " ".join([word for word in jieba.cut(x)
if word not in stop_words and x]))
data['content'] = data['content'].apply(lambda x: " ".join([word for word in jieba.cut(str(x))
if word not in stop_words and x]))
articles=[]
for title,summary,content in zip(data['title'].tolist(),
data['summary'].tolist(),
data['content'].tolist()):
article=title+summary+content
articles.append(article)
with open(filename,'wb') as out_data:
pickle.dump(articles,out_data,pickle.HIGHEST_PROTOCOL)
articles = [split_artilce(article) for article in articles]
return articles
def train_doc2vec(articles):
tag_tokenized=[gensim.models.doc2vec.TaggedDocument(articles[i],[i]) for i in range(len(articles))]
model =Doc2Vec(size=200, min_count=2, iter=200)
model.build_vocab(tag_tokenized)
model.train(tag_tokenized, total_examples=model.corpus_count, epochs=model.iter)
# 保存模型
model.save('model/doc2vec.model')
def most_sim_docs(articles):
"""
计算文本之间的相似度
测试训练的模型， 输出相似文章
下面做了简单测试，假如我们测试下articles第一个doc相似的10篇文章，然后我们可以看到原文相似度最高
:return:
"""
model_dm = Doc2Vec.load("model/doc2vec.model")
test_text=articles[random.randint(1,1800)]
print(test_text)
inferred_vector_dm = model_dm.infer_vector(test_text)
# print(inferred_vector_dm)
sims = model_dm.docvecs.most_similar([inferred_vector_dm], topn=10)
for index, sim in sims:
sentence = articles[index]
words = ''
for word in sentence[0]:
words = words + word + ' '
print(sim,articles[index])
def cluster(articles):
"""
doc2vec + kmeans 做文本聚类
:return:
"""
articles = [gensim.models.doc2vec.TaggedDocument(articles[i], [i]) for i in range(len(articles))]
infered_vectors_list=[]
print("load doc2vec model...")
model_dm=Doc2Vec.load('model/doc2vec.model')
print("load train vectors...")
for text,label in articles:
vector=model_dm.infer_vector(text)
infered_vectors_list.append(vector)
print("train k-mean model ...")
kmean_model=KMeans(n_clusters=8)
kmean_model.fit(infered_vectors_list)
labels = kmean_model.predict(infered_vectors_list[0:1800])
cluster_centers = kmean_model.cluster_centers_
with open('model/classification.txt','w',encoding='utf-8') as out_f:
for i in range(1800):
string = ""
text = articles[i][0]
for word in text:
string = string + word
string = string + '\t'
string = string + str(labels[i])
string = string + '\n'
out_f.write(string)
if __name__ == '__main__':
articles = load_articles()
# train_doc2vec(articles)
most_sim_docs(articles)
# cluster(articles)