tensorflow + LSTM 实现短文本评论情感分类

深藏阁楼爱情的钟 2021-11-05 09:20 411阅读 0赞

# 1 文本数据介绍及来源 #

本文使用的数据集为网站爬取的短文本评论，其中由于数据量较少，不区分验证集与测试集，并按照4：1的比例将原始有标签数据分为训练集与测试集。  
原始数据文件压缩包网盘链接：[https://pan.baidu.com/s/13vwd3lfKWfXlD1a8uB6ngg][https_pan.baidu.com_s_13vwd3lfKWfXlD1a8uB6ngg]  
提取码：urj2  
注：解压后的text\_data文件夹放置在与程序的同级目录下。  
原始数据以txt文本格式保存，标签分为pos与neg，因此预处理主要是将标签与数据分开对应保存，并将数据分为训练集与测试集。

# 2 模型网络结构 #

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl80MzQ4MzM4MQ_size_19_color_FFFFFF_t_70][]

如上图所示为LSTM进行本文所示的短文本情感分类的原理示意图。  
其中，对于每一条短文本样本而言，里面每一个字Wt的矩阵形状为(,1,60)，其中第一个维度表示样本数，第二个维度值为1表示一个字，第三个维度用长为60的向量表示这个字Wt对应的embedding\_vector。  
在网络超参数中，num\_hidden\_units 即LSTM细胞隐藏层大小设置为256，即图中使用黄色矩形代表的softmax层，具体的输入输出关系如图所示。最终我们获得整个所有cell的Ot以及最后一个cell的state(此state后续并未使用)。

# 3 数据预处理及字库构建 #

import os
    # 获取text_data文件夹下的所有文件路径
    temp_list = list(os.walk(r"text_data"))
    original = temp_list[0][0]
    file_name = temp_list[0][2]
    path_list = [original + "\\" + eve_name for eve_name in file_name]
    # 创建所需文件
    train_data = open(r"train_data.txt", "w", encoding="utf-8")
    train_label = open(r"train_label.txt", "w", encoding="utf-8")
    test_data = open(r"test_data.txt", "w", encoding="utf-8")
    test_label = open(r"test_label.txt", "w", encoding="utf-8")
    vocabulary = open(r"vocabulary.txt", "w", encoding="utf-8")
    # 将原始数据进行标签分离与训练测试集分离
    for every_path in path_list:
        with open(every_path, "r", encoding="utf-8") as temp_file:
            corpus = [eve for eve in temp_file if len(eve.strip("\n")) != 0]
            limit1 = len(corpus)*0.9
            limit2 = len(corpus)*0.1
            for i in range(len(corpus)):
                if limit2 < i < limit1:
                    if corpus[i][:3] == "pos":
                        train_data.write(corpus[i][3:])
                        train_label.write("1" + "\n")
                    else:
                        train_data.write(corpus[i][3:])
                        train_label.write("0" + "\n")
                else:
                    if corpus[i][:3] == "pos":
                        test_data.write(corpus[i][3:])
                        test_label.write("1" + "\n")
                    else:
                        test_data.write(corpus[i][3:])
                        test_label.write("0" + "\n")
    # 创建字库vocabulary，包含原始数据中所有的字，写入vocabulary.txt待用
    with open(r"test_data.txt", "r", encoding="utf-8") as file1:
    	corpus1 = [eve for eve in file1]
    with open(r"train_data.txt", "r", encoding="utf-8") as file2:
    	corpus2 = [eve for eve in file2]
    with open(r"vocabulary.txt","w",encoding="utf-8") as file3:
        word_list = []
        corpus = corpus1 + corpus2
        for line in corpus:
            for word in line:
                word_list.append(word)
        word_list = list(set(word_list))
        for word in word_list:
            file3.write(word + "\n")

# 4 tensorflow + LSTM实现文本情感二分类 #

按照上图所示模型，具体程序如下所示。

with open(r"train_data.txt", "r", encoding="utf-8") as file1:
        corpus = [eve.strip("\n") for eve in file1]
    with open(r"vocabulary.txt", "r", encoding="utf-8") as file2:
        vocabulary = [word.strip("\n") for word in file2]
    with open(r"train_label.txt", "r", encoding="utf-8") as file3:
        label_list = [int(eve.strip("\n")) for eve in file3]
    assert len(label_list) == len(corpus)
    
    word2id = { word:id_ for id_, word in enumerate(vocabulary)}
    line2id = lambda line: [word2id[word] for word in line]  # 将一句话转化为id的列表list
    train_list = [line2id(line) for line in corpus]
    
    import tensorflow.contrib.keras as kr
    train_x = kr.preprocessing.sequence.pad_sequences(train_list, 100)  # 长度一致train_x
    train_y = kr.utils.to_categorical(label_list, num_classes=2)
    import tensorflow as tf
    tf.compat.v1.reset_default_graph()
    X_holder = tf.compat.v1.placeholder(tf.int32, [None, 100])  # 占位
    Y_holder = tf.compat.v1.placeholder(tf.float32, [None, 2])
    # 做词嵌入工作 注意2775是自由生成的行向量，必须大于等于实际的数据
    embedding = tf.compat.v1.get_variable('embedding', [2775, 60])  # 一种初始化变量的方法，随机初始化了矩阵变量
    embedding_inputs = tf.nn.embedding_lookup(embedding, X_holder)
    num_hidden_units = 256  # LSTM细胞隐藏层大小
    num_fc1_units = 64 # 第1个全连接下一层的大小
    dropout_keep_probability = 0.5  # dropout保留比例
    learning_rate = 1e-2 # 学习率
    # 整体结构：输入-LSTM-全连接层-dropout-全连接层-二分类softmax层
    gru_cell = tf.contrib.rnn.BasicLSTMCell(num_hidden_units)  # 实例化一个LSTM单元，此处也可以直接替换为GRU单元
    outputs, _ = tf.nn.dynamic_rnn(gru_cell,
                                       embedding_inputs,
                                       dtype=tf.float32)  # _为最后一个cell的state
    last_cell = outputs[:, -1, :]  # 取出最后一个cell的outputs做输出，多对一
    full_connect1 = tf.layers.dense(last_cell,
                                    num_fc1_units)
    full_connect1_dropout = tf.contrib.layers.dropout(full_connect1,
                                                      dropout_keep_probability)
    full_connect1_activate = tf.nn.relu(full_connect1_dropout)
    full_connect2 = tf.layers.dense(full_connect1_activate, 2)  # 连接二分类器
    predict_Y = tf.nn.softmax(full_connect2)  # 激活
    
    cross_entropy = tf.nn.softmax_cross_entropy_with_logits_v2(labels=Y_holder, logits=full_connect2)
    loss = tf.reduce_mean(cross_entropy)  # 熵的平均值
    optimizer = tf.train.AdamOptimizer(learning_rate)  # 定义优化器
    train = optimizer.minimize(loss)  # 将优化器与损失值连接起来
    
    isCorrect = tf.equal(tf.argmax(Y_holder, 1), tf.argmax(predict_Y, 1))  # 判断是否正确
    accuracy = tf.reduce_mean(tf.cast(isCorrect, tf.float32))  # 判断准确率
    
    init = tf.global_variables_initializer()  # 初始化参数
    session = tf.Session()
    session.run(init)
    
    # 读取测试集进行测试
    with open(r"test_data.txt", "r", encoding="utf-8") as file4:
        corpus_ = [eve.strip("\n") for eve in file4]
    with open(r"test_label.txt", "r", encoding="utf-8") as file5:
        label_list_ = [int(eve.strip("\n")) for eve in file5]
    assert len(label_list_) == len(corpus_)
    test_list = [line2id(line) for line in corpus_]
    test_x = kr.preprocessing.sequence.pad_sequences(test_list, 100)  # 长度一致train_x
    test_y = kr.utils.to_categorical(label_list_, num_classes=2)
    import random
    for i in range(3000):
        selected_index = random.sample(list(range(len(train_y))), k=60)  # 批训练
        batch_X = train_x[selected_index]
        batch_Y = train_y[selected_index]
        session.run(train, { X_holder:batch_X, Y_holder:batch_Y})
        step = i + 1
        if step % 100 == 0:
            selected_index = random.sample(list(range(len(test_y))), k=150)
            batch_X = test_x[selected_index]
            batch_Y = test_y[selected_index]
            loss_value, accuracy_value = session.run([loss, accuracy], { X_holder:batch_X, Y_holder:batch_Y})
            print('step:%d loss:%.4f accuracy:%.4f' %(step, loss_value, accuracy_value))

# 6 运行结果 #

step:100 loss:0.4196 accuracy:0.8400
    step:200 loss:0.2480 accuracy:0.9000
    step:300 loss:0.2244 accuracy:0.9200
    step:400 loss:0.1739 accuracy:0.9333
    step:500 loss:0.3256 accuracy:0.9000
    step:600 loss:0.3013 accuracy:0.9000
    step:700 loss:0.1742 accuracy:0.9333
    step:800 loss:0.3055 accuracy:0.9000
    step:900 loss:0.2212 accuracy:0.9067
    step:1000 loss:0.3092 accuracy:0.8800
    step:1100 loss:0.2734 accuracy:0.9000
    step:1200 loss:0.2690 accuracy:0.8733
    step:1300 loss:0.2409 accuracy:0.9067
    step:1400 loss:0.3447 accuracy:0.8800
    step:1500 loss:0.2810 accuracy:0.8933
    step:1600 loss:0.2851 accuracy:0.9200
    step:1700 loss:0.2294 accuracy:0.9333
    step:1800 loss:0.2828 accuracy:0.8933
    step:1900 loss:0.3330 accuracy:0.8933
    step:2000 loss:0.3764 accuracy:0.9200
    step:2100 loss:0.1498 accuracy:0.9200
    step:2200 loss:0.1861 accuracy:0.9467
    step:2300 loss:0.2425 accuracy:0.9133
    step:2400 loss:0.2472 accuracy:0.9333
    step:2500 loss:0.3103 accuracy:0.8933
    step:2600 loss:0.2019 accuracy:0.9200
    step:2700 loss:0.1626 accuracy:0.9333
    step:2800 loss:0.5135 accuracy:0.9200
    step:2900 loss:0.3478 accuracy:0.9000
    step:3000 loss:0.3352 accuracy:0.9133

[https_pan.baidu.com_s_13vwd3lfKWfXlD1a8uB6ngg]: https://pan.baidu.com/s/13vwd3lfKWfXlD1a8uB6ngg
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl80MzQ4MzM4MQ_size_19_color_FFFFFF_t_70]: /images/20211104/8d2c26e30f9f4fa1a1cec29170affc02.png