Keras下的文本情感分析简介。与MLP,RNN,LSTM模型下的文本情感测试

# coding: utf-8

# In[1]:


import urllib.request
import os
import tarfile


# In[2]:


url="http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz"
filepath="example/data/aclImdb_v1.tar.gz"
if not os.path.isfile(filepath):
    result=urllib.request.urlretrieve(url,filepath)
    print('downloaded:',result)
if not os.path.exists("example/data/aclImdb_v1/aclImdb"):
    tfile = tarfile.open("data/aclImdb_v1.tar.gz", 'r:gz')
    result=tfile.extractall('data/')


# In[3]:


from keras.datasets import imdb
from keras.preprocessing import sequence
from keras.preprocessing.text import Tokenizer


# In[4]:


import re
def rm_tags(text):
    re_tag = re.compile(r'<[^>]+>')
    return re_tag.sub('', text)


# In[5]:


import os
def read_files(filetype):
    path = "example/data/aclImdb_v1/aclImdb/"
    file_list=[]

    positive_path=path + filetype+"/pos/"
    for f in os.listdir(positive_path):
        file_list+=[positive_path+f]
    
    negative_path=path + filetype+"/neg/"
    for f in os.listdir(negative_path):
        file_list+=[negative_path+f]
        
    print('read',filetype, 'files:',len(file_list))
    all_labels = ([1] * 12500 + [0] * 12500) 
    
    all_texts  = []
    for fi in file_list:
        with open(fi,encoding='utf8') as file_input:
            all_texts += [rm_tags(" ".join(file_input.readlines()))]
            
    return all_labels,all_texts


# In[6]:


y_train,train_text=read_files("train")


# In[7]:


y_test,test_text=read_files("test")


# In[8]:


train_text[0]


# In[9]:


y_train[0]


# In[10]:


train_text[12500]


# In[11]:


y_train[12500]


# In[12]:


token = Tokenizer(num_words=2000)
token.fit_on_texts(train_text)


# In[13]:


print(token.document_count)
print(token.word_index)


# In[14]:


x_train_seq = token.texts_to_sequences(train_text)
x_test_seq  = token.texts_to_sequences(test_text)


# In[15]:


print(x_train_seq[0])


# In[16]:


x_train = sequence.pad_sequences(x_train_seq, maxlen=100)
x_test  = sequence.pad_sequences(x_test_seq,  maxlen=100)


# In[17]:


x_train[0]


# In[18]:


from keras.models import Sequential
from keras.layers.core import Dense, Dropout, Activation,Flatten
from keras.layers.embeddings import Embedding
model = Sequential()
model.add(Embedding(output_dim=32,
                    input_dim=2000, 
                    input_length=100))
model.add(Dropout(0.2))
model.add(Flatten())
model.add(Dense(units=256,
                activation='relu' ))
model.add(Dropout(0.2))
model.add(Dense(units=1,
                activation='sigmoid' ))
model.summary()


# In[19]:


model.compile(loss='binary_crossentropy', 
              optimizer='adam', 
              metrics=['accuracy'])
train_history =model.fit(x_train, y_train,batch_size=100, 
                         epochs=10,verbose=2,
                         validation_split=0.2)


# In[20]:


get_ipython().magic('pylab inline')
import matplotlib.pyplot as plt
def show_train_history(train_history,train,validation):
    plt.plot(train_history.history[train])
    plt.plot(train_history.history[validation])
    plt.title('Train History')
    plt.ylabel(train)
    plt.xlabel('Epoch')
    plt.legend(['train', 'validation'], loc='upper left')
    plt.show()


# In[21]:


show_train_history(train_history,'acc','val_acc')
show_train_history(train_history,'loss','val_loss')


# In[22]:


scores = model.evaluate(x_test, y_test, verbose=1)
scores[1]


# In[23]:


probility=model.predict(x_test)


# In[24]:


probility[:10]


# In[25]:


probility[12500:12510]


# In[26]:


predict=model.predict_classes(x_test)


# In[27]:


predict_classes=predict.reshape(-1)


# In[28]:


SentimentDict={1:'正面的',0:'负面的'}
def display_test_Sentiment(i):
    print(test_text[i])
    print('标签label:',SentimentDict[y_test[i]],
          '预测结果:',SentimentDict[predict_classes[i]])


# In[29]:


display_test_Sentiment(2)


# In[30]:


display_test_Sentiment(12505)


# In[31]:


from keras.models import Sequential
from keras.layers.core import Dense, Dropout, Activation
from keras.layers.embeddings import Embedding
from keras.layers.recurrent import SimpleRNN
model = Sequential()
model.add(Embedding(output_dim=32,
                    input_dim=2000, 
                    input_length=100))
model.add(Dropout(0.35))
model.add(SimpleRNN(units=16))
model.add(Dense(units=256,activation='relu' ))
model.add(Dropout(0.35))
model.add(Dense(units=1,activation='sigmoid' ))
model.summary()


# In[32]:


model.compile(loss='binary_crossentropy', 
              optimizer='adam', 
              metrics=['accuracy'])
train_history =model.fit(x_train, y_train,batch_size=100, 
                         epochs=10,verbose=2,
                         validation_split=0.2)


# In[33]:


scores = model.evaluate(x_test, y_test, verbose=1)
scores[1]


# In[34]:


from keras.models import Sequential
from keras.layers.core import Dense, Dropout, Activation,Flatten
from keras.layers.embeddings import Embedding
from keras.layers.recurrent import LSTM
model = Sequential()
model.add(Embedding(output_dim=32,
                    input_dim=2000, 
                    input_length=100))
model.add(Dropout(0.2))
model.add(LSTM(32))
model.add(Dense(units=256,
                activation='relu' ))
model.add(Dropout(0.2))
model.add(Dense(units=1,
                activation='sigmoid' ))
model.summary()


# In[35]:


model.compile(loss='binary_crossentropy', 
              #optimizer='rmsprop', 
              optimizer='adam', 
              metrics=['accuracy'])
train_history =model.fit(x_train, y_train,batch_size=100, 
                         epochs=10,verbose=2,
                         validation_split=0.2)


# In[36]:


show_train_history(train_history,'acc','val_acc')
show_train_history(train_history,'loss','val_loss')
scores = model.evaluate(x_test, y_test, verbose=1)
scores[1]


# In[ ]:

 文本来源于IMDb网络电影数据集。下载，放到合适的路径下，然后，开始。

过滤掉HTML标签。因为数据集中有相关标签。：

之后读取所有数据和目标标签，然后建立字典：

将文本转化为数字串：

格式化数字串长度为100

建立MLP模型，其中嵌入层将每个长度为100的数字串转为100个32维的向量，将文字映射成多维的几何空间向量，让每一个文字有上下的关联性。

编译，训练，绘图，评估后的准确率：

建立RNN模型，有关RNN模型的介绍：https://www.cnblogs.com/bai2018/p/10466418.html

 

测试评估：

建立LSTM模型，相关介绍：https://www.cnblogs.com/bai2018/p/10466497.html

 准确率：