基于word2vec的分类任务

# bs4 nltk gensim
import os
import re
import numpy as np
import pandas as pd

from bs4 import BeautifulSoup

from sklearn.feature_extraction.text import CountVectorizer
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import confusion_matrix
from sklearn.linear_model import LogisticRegression
import nltk
from nltk.corpus import stopwords

  

 

  
1
  
2
  
3
  
4
  
5
  
6
  
7
  
8
  
9
  
10
  
11
  
12
  
13
  
14
 

用pandas读入训练数据

df = pd.read_csv('../data/labeledTrainData.tsv', sep='\t', escapechar='\\')
print('Number of reviews: {}'.format(len(df)))
df.head()

  

 

  
1
  
2
  
3
 

Number of reviews: 25000

  

 

  
1
 

对影评数据做预处理，大概有以下环节：

1. 去掉html标签
2. 移除标点
3. 切分成词/token
4. 去掉停用词
5. 重组为新的句子

df['review'][1000]

  

 

  
1
 

"I watched this movie really late last night and usually if it's late then I'm pretty forgiving of movies. Although I tried, I just could not stand this movie at all, it kept 
getting worse and worse as the movie went on. Although I know it's suppose to be a comedy but I didn't find it very funny. It was also an especially unrealistic, and 
jaded portrayal of rural life. In case this is what any of you think country life is like, it's definitely not. I do have to agree that some of the guy cast members were cute, but the french guy was really fake. I do have to agree that it tried to have a good lesson in the story, but overall my recommendation is that no one over 8 
watch it, it's just too annoying."

  

 

  
1
  
2
  
3
  
4
 

#去掉HTML标签的数据
example = BeautifulSoup(df['review'][1000], 'html.parser').get_text()
example

  

 

  
1
  
2
  
3
 

"I watched this movie really late last night and usually if it's late then I'm pretty forgiving of movies. Although I tried, I just could not stand this movie at all, it kept 
getting worse and worse as the movie went on. Although I know it's suppose to be a comedy but I didn't find it very funny. It was also an especially unrealistic, and jaded portrayal of rural life. In case this is what any of you think country life is like, it's definitely not. I do have to agree that some of the guy cast members were 
cute, but the french guy was really fake. I do have to agree that it tried to have a good lesson in the story, but overall my recommendation is that no one over 8
 watch it, it's just too annoying."

  

 

  
1
  
2
  
3
  
4
 

#去掉标点符号
example_letters = re.sub(r'[^a-zA-Z]', ' ', example)
example_letters

  

 

  
1
  
2
  
3
 

'I watched this movie really late last night and usually if it s late then I m pretty forgiving of movies  Although I tried  I just could not stand this movie at all  it kept 
getting worse and worse as the movie went on  Although I know it s suppose to be a comedy but I didn t find it very funny  It was also an especially unrealistic  
and jaded portrayal of rural life  In case this is what any of you think country life is like  it s definitely not  I do have to agree that some of the guy cast members 
were cute  but the french guy was really fake  I do have to agree that it tried to have a good lesson in the story  but overall my recommendation is that no one 
over   watch it  it s just too annoying '

  

 

  
1
  
2
  
3
  
4
  
5
 

words = example_letters.lower().split()
words

  

 

  
1
  
2
  
3
 

['i',
 'watched',
 'this',
 'movie',
 'really',
 'late',
 'last',
 'night',
 'and',
 'usually',
 'if',
 'it',
 's',
 'late',
 'then',
 'i',
 'm',
 'pretty',
 'forgiving',
 'of',
 'movies',
 'although',
 'i',
 'tried',
 'i',
 'just',
 'could',
 'not',
 'stand',
 'this',
 'movie',
 'at',
 'all',
 'it',
 'kept',
 'getting',
 'worse',
 'and',
 'worse',
 'as',
 'the',
 'movie',
 'went',
 'on',
 'although',
 'i',
 'know',
 'it',
 's',
 'suppose',
 'to',
 'be',
 'a',
 'comedy',
 'but',
 'i',
 'didn',
 't',
 'find',
 'it',
 'very',
 'funny',
 'it',
 'was',
 'also',
 'an',
 'especially',
 'unrealistic',
 'and',
 'jaded',
 'portrayal',
 'of',
 'rural',
 'life',
 'in',
 'case',
 'this',
 'is',
 'what',
 'any',
 'of',
 'you',
 'think',
 'country',
 'life',
 'is',
 'like',
 'it',
 's',
 'definitely',
 'not',
 'i',
 'do',
 'have',
 'to',
 'agree',
 'that',
 'some',
 'of',
 'the',
 'guy',
 'cast',
 'members',
 'were',
 'cute',
 'but',
 'the',
 'french',
 'guy',
 'was',
 'really',
 'fake',
 'i',
 'do',
 'have',
 'to',
 'agree',
 'that',
 'it',
 'tried',
 'to',
 'have',
 'a',
 'good',
 'lesson',
 'in',
 'the',
 'story',
 'but',
 'overall',
 'my',
 'recommendation',
 'is',
 'that',
 'no',
 'one',
 'over',
 'watch',
 'it',
 'it',
 's',
 'just',
 'too',
 'annoying']

  

 

  
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
  
128
  
129
  
130
  
131
  
132
  
133
  
134
  
135
  
136
  
137
  
138
  
139
  
140
  
141
  
142
  
143
  
144
 

#去停用词
stopwords = {}.fromkeys([ line.rstrip() for line in open('../stopwords.txt')])
words_nostop = [w for w in words if w not in stopwords]
words_nostop

  

 

  
1
  
2
  
3
  
4
 

['watched',
 'movie',
 'late',
 'night',
 'late',
 'pretty',
 'forgiving',
 'movies',
 'stand',
 'movie',
 'worse',
 'worse',
 'movie',
 'suppose',
 'comedy',
 'didn',
 'funny',
 'unrealistic',
 'jaded',
 'portrayal',
 'rural',
 'life',
 'country',
 'life',
 'agree',
 'guy',
 'cast',
 'cute',
 'french',
 'guy',
 'fake',
 'agree',
 'lesson',
 'story',
 'recommendation',
 'watch',
 'annoying']

  

 

  
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
 

eng_stopwords = set(stopwords)

def clean_text(text): text = BeautifulSoup(text, 'html.parser').get_text() text = re.sub(r'[^a-zA-Z]', ' ', text) words = text.lower().split() words = [w for w in words if w not in eng_stopwords] return ' '.join(words)

  

 

  
1
  
2
  
3
  
4
  
5
  
6
  
7
  
8
 

df['review'][1000]

  

 

  
1
 

"I watched this movie really late last night and usually if it's late then I'm pretty forgiving of movies. Although I tried, I just could not stand this movie at all, it kept
 getting worse and worse as the movie went on. Although I know it's suppose to be a comedy but I didn't find it very funny. It was also an especially unrealistic, 
 and jaded portrayal of rural life. In case this is what any of you think country life is like, it's definitely not. I do have to agree that some of the guy cast members 
 were cute, but the french guy was really fake. I do have to agree that it tried to have a good lesson in the story, but overall my recommendation is that no one 
 over 8 watch it, it's just too annoying."

  

 

  
1
  
2
  
3
  
4
  
5
 

clean_text(df['review'][1000])

  

 

  
1
 

'watched movie late night late pretty forgiving movies stand movie worse worse movie suppose comedy didn funny unrealistic jaded portrayal rural life country 
life agree guy cast cute french guy fake agree lesson story recommendation watch annoying'

  

 

  
1
  
2
 

清洗数据添加到dataframe里

df['clean_review'] = df.review.apply(clean_text)
df.head()

  

 

  
1
  
2
 

抽取bag of words特征(用sklearn的CountVectorizer)

vectorizer = CountVectorizer(max_features = 5000) 
train_data_features = vectorizer.fit_transform(df.clean_review).toarray()
train_data_features.shape

  

 

  
1
  
2
  
3
 

(25000, 5000)

  

 

  
1
 

from sklearn.cross_validation import train_test_split
X_train, X_test, y_train, y_test = train_test_split(train_data_features,df.sentiment,test_size = 0.2, random_state = 0)

  

 

  
1
  
2
 

import matplotlib.pyplot as plt
import itertools
def plot_confusion_matrix(cm, classes, title='Confusion matrix', cmap=plt.cm.Blues): """ This function prints and plots the confusion matrix. """ plt.imshow(cm, interpolation='nearest', cmap=cmap) plt.title(title) plt.colorbar() tick_marks = np.arange(len(classes)) plt.xticks(tick_marks, classes, rotation=0) plt.yticks(tick_marks, classes) thresh = cm.max() / 2. for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])): plt.text(j, i, cm[i, j], horizontalalignment="center", color="white" if cm[i, j] > thresh else "black") plt.tight_layout() plt.ylabel('True label') plt.xlabel('Predicted label')

  

 

  
1
  
2
  
3
  
4
  
5
  
6
  
7
  
8
  
9
  
10
  
11
  
12
  
13
  
14
  
15
  
16
  
17
  
18
  
19
  
20
  
21
  
22
  
23
  
24
 

训练分类器

LR_model = LogisticRegression()
LR_model = LR_model.fit(X_train, y_train)
y_pred = LR_model.predict(X_test)
cnf_matrix = confusion_matrix(y_test,y_pred)

print("Recall metric in the testing dataset: ", cnf_matrix[1,1]/(cnf_matrix[1,0]+cnf_matrix[1,1]))

print("accuracy metric in the testing dataset: ", (cnf_matrix[1,1]+cnf_matrix[0,0])/(cnf_matrix[0,0]+cnf_matrix[1,1]+cnf_matrix[1,0]+cnf_matrix[0,1]))

# Plot non-normalized confusion matrix
class_names = [0,1]
plt.figure()
plot_confusion_matrix(cnf_matrix , classes=class_names , title='Confusion matrix')
plt.show()

  

 

  
1
  
2
  
3
  
4
  
5
  
6
  
7
  
8
  
9
  
10
  
11
  
12
  
13
  
14
  
15
  
16
 

Recall metric in the testing dataset:  0.853181076672
accuracy metric in the testing dataset:  0.8454

  

 

  
1
  
2
 

df = pd.read_csv('../data/unlabeledTrainData.tsv', sep='\t', escapechar='\\')
print('Number of reviews: {}'.format(len(df)))
df.head()

  

 

  
1
  
2
  
3
 

Number of reviews: 50000

  

 

  
1
 

df['clean_review'] = df.review.apply(clean_text)
df.head()

  

 

  
1
  
2
 

review_part = df['clean_review']
review_part.shape

  

 

  
1
  
2
 

(50000,)

  

 

  
1
 

import warnings
warnings.filterwarnings("ignore")

tokenizer = nltk.data.load('tokenizers/punkt/english.pickle')


def split_sentences(review): raw_sentences = tokenizer.tokenize(review.strip()) sentences = [clean_text(s) for s in raw_sentences if s] return sentences
sentences = sum(review_part.apply(split_sentences), [])
print('{} reviews -> {} sentences'.format(len(review_part), len(sentences)))

  

 

  
1
  
2
  
3
  
4
  
5
  
6
  
7
  
8
  
9
  
10
  
11
  
12
 

50000 reviews -> 50000 sentences

  

 

  
1
 

sentences[0]

  

 

  
1
 

'watching time chasers obvious bunch friends sitting day film school hey pool money bad movie bad movie dull story bad script lame acting poor cinematography 
bottom barrel stock music corners cut prevented film release life'

  

 

  
1
  
2
 

sentences_list = []
for line in sentences: sentences_list.append(nltk.word_tokenize(line))

  

 

  
1
  
2
  
3
 

• sentences：可以是一个list

• sg： 用于设置训练算法，默认为0，对应CBOW算法；sg=1则采用skip-gram算法。

• size：是指特征向量的维度，默认为100。大的size需要更多的训练数据,但是效果会更好. 推荐值为几十到几百。

• window：表示当前词与预测词在一个句子中的最大距离是多少

• alpha: 是学习速率

• seed：用于随机数发生器。与初始化词向量有关。

• min_count: 可以对字典做截断. 词频少于min_count次数的单词会被丢弃掉, 默认值为5

• max_vocab_size: 设置词向量构建期间的RAM限制。如果所有独立单词个数超过这个，则就消除掉其中最不频繁的一个。每一千万个单词需要大约1GB的RAM。设置成None则没有限制。

• workers参数控制训练的并行数。

• hs: 如果为1则会采用hierarchica·softmax技巧。如果设置为0（defau·t），则negative sampling会被使用。

• negative: 如果>0,则会采用negativesamp·ing，用于设置多少个noise words

• iter： 迭代次数，默认为5

# 设定词向量训练的参数
num_features = 300 # Word vector dimensionality
min_word_count = 40   # Minimum word count
num_workers = 4 # Number of threads to run in parallel
context = 10 # Context window size
model_name = '{}features_{}minwords_{}context.model'.format(num_features, min_word_count, context)

  

 

  
1
  
2
  
3
  
4
  
5
  
6
 

from gensim.models.word2vec import Word2Vec
model = Word2Vec(sentences_list, workers=num_workers, \ size=num_features, min_count = min_word_count, \ window = context)

# If you don't plan to train the model any further, calling 
# init_sims will make the model much more memory-efficient.
model.init_sims(replace=True)

# It can be helpful to create a meaningful model name and 
# save the model for later use. You can load it later using Word2Vec.load()
model.save(os.path.join('..', 'models', model_name))

  

 

  
1
  
2
  
3
  
4
  
5
  
6
  
7
  
8
  
9
  
10
  
11
  
12
 

print(model.doesnt_match(['man','woman','child','kitchen']))
#print(model.doesnt_match('france england germany berlin'.split())

  

 

  
1
  
2
 

kitchen

  

 

  
1
 

model.most_similar("boy")

  

 

  
1
 

[('girl', 0.7018299698829651),
 ('astro', 0.6647905707359314),
 ('teenage', 0.6317306160926819),
 ('frat', 0.60948246717453),
 ('dad', 0.6011481285095215),
 ('yr', 0.6010577082633972),
 ('teenager', 0.5974895358085632),
 ('brat', 0.5941195487976074),
 ('joshua', 0.5832049250602722),
 ('father', 0.5825375914573669)]

  

 

  
1
  
2
  
3
  
4
  
5
  
6
  
7
  
8
  
9
  
10
 

model.most_similar("bad")

  

 

  
1
 

[('worse', 0.7071679830551147),
 ('horrible', 0.7065873742103577),
 ('terrible', 0.6872220635414124),
 ('sucks', 0.6666240692138672),
 ('crappy', 0.6634873747825623),
 ('lousy', 0.6494461297988892),
 ('horrendous', 0.6371070742607117),
 ('atrocious', 0.62550288438797),
 ('suck', 0.6224384307861328),
 ('awful', 0.619296669960022)]

  

 

  
1
  
2
  
3
  
4
  
5
  
6
  
7
  
8
  
9
  
10
 

df = pd.read_csv('../data/labeledTrainData.tsv', sep='\t', escapechar='\\')
df.head()

  

 

  
1
  
2
 

from nltk.corpus import stopwords
eng_stopwords = set(stopwords.words('english'))

def clean_text(text, remove_stopwords=False): text = BeautifulSoup(text, 'html.parser').get_text() text = re.sub(r'[^a-zA-Z]', ' ', text) words = text.lower().split() if remove_stopwords: words = [w for w in words if w not in eng_stopwords] return words

def to_review_vector(review): global word_vec review = clean_text(review, remove_stopwords=True) #print (review) #words = nltk.word_tokenize(review) word_vec = np.zeros((1,300)) for word in review: #word_vec = np.zeros((1,300)) if word in model: word_vec += np.array([model[word]]) #print (word_vec.mean(axis = 0)) return pd.Series(word_vec.mean(axis = 0))

train_data_features = df.review.apply(to_review_vector)
train_data_features.head()

  

 

  
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
 

5 rows × 300 columns

from sklearn.cross_validation import train_test_split

X_train, X_test, y_train, y_test = train_test_split(train_data_features,df.sentiment,test_size = 0.2, random_state = 0)

  

 

  
1
  
2
  
3
 

LR_model = LogisticRegression()
LR_model = LR_model.fit(X_train, y_train)
y_pred = LR_model.predict(X_test)
cnf_matrix = confusion_matrix(y_test,y_pred)

print("Recall metric in the testing dataset: ", cnf_matrix[1,1]/(cnf_matrix[1,0]+cnf_matrix[1,1]))

print("accuracy metric in the testing dataset: ", (cnf_matrix[1,1]+cnf_matrix[0,0])/(cnf_matrix[0,0]+cnf_matrix[1,1]+cnf_matrix[1,0]+cnf_matrix[0,1]))

# Plot non-normalized confusion matrix
class_names = [0,1]
plt.figure()
plot_confusion_matrix(cnf_matrix , classes=class_names , title='Confusion matrix')
plt.show()

  

 

  
1
  
2
  
3
  
4
  
5
  
6
  
7
  
8
  
9
  
10
  
11
  
12
  
13
  
14
  
15
  
16
 

Recall metric in the testing dataset:  0.87969004894
accuracy metric in the testing dataset:  0.865

  

 

  
1
  
2
 

文章来源: maoli.blog.csdn.net，作者：刘润森！，版权归原作者所有，如需转载，请联系作者。

原文链接：maoli.blog.csdn.net/article/details/89462163