Tensorflow+gensim实现文章自动审核功能

首先，一定要吐槽一下，千万不要拿笔记本跑训练集很大的模型，不然真的能跑到吐血三升，而且电脑卡到宕机之后好不容易等到电脑能动了就是报错“内存不足”，简直能逼死我。每次跑程序都得把虚存开到最大，然后看着我的电脑在那卡卡卡，我都在想会不会把我电脑跑废了。当然结局是美好的，因为模型跑出来的审核结果正确率在94%以上，简直是惊喜到不行，嗯，基于这个理由，给我配工作站让我专门跑算法了，开心~

好了，言归正传，由于THUCTC模型对于文章审核方面的正确率很低，可能是由于切词结果和词的权重不合适的关系，导致我需要去寻找新的合适的模型和算法，最后选择了谷歌的gensim（主要针对自然语言的处理，里面也包括tf-idf）构建和训练词向量，神经网络选用了CNN卷积神经网络，分词选用的jieba分词，按照词性筛选需要的关键词构建词向量。比较核心的就是这几个了。还有一些对文档的处理，等到代码的时候就知道了，主要是筛除一些干扰的无用的词或标点，将文章统一长度等

话不多说，让我们来看看代码吧。

首先就是训练文档的准备，以什么形式不重要，只要后续能够读取做成模型的参数就行，我这里选用的是tensorflow的文本格式.tfRecords。如果是用普通文本的就自己读写文件就好了，我这里就以我用的tfRecords的存取来说明了。

代码如下，代码中也包括对于预测文档的准备及粗处理：

class loadData: 
    # =============================================================================
    # 从配置文件中读取信息start
    # =============================================================================
    def __init__(self, base_dir = '.', path="database.conf"):
            self.base_dir = base_dir
            self.path = path
            self.cf = configparser.ConfigParser()
            self.cf.read(self.path)

    def get_configInfo(self, field, key):
        result = ""
        try:
            result = self.cf.get(field, key)
        except:
            result = ""
        return result

    def read_config(self, config_file_path="database.conf", field="db"):
        cf = configparser.ConfigParser()
        try:
            cf.read(config_file_path)
            db_host = cf.get(field, "db_host")
            db_user = cf.get(field, "db_user")
            db_pass = cf.get(field, "db_pass")
            db_name = cf.get(field, "db_name")
            db_charset = cf.get(field, "db_charset")
            db_port = cf.get(field, "db_port")
        except Exception as inst:
            traceback.print_exc()
            print(type(inst))
            print(inst.args)
            print(inst)
        return db_host,db_user,db_pass,db_name,db_charset,db_port
    # =============================================================================
    # 从配置文件中读取信息end
    # =============================================================================

    # =============================================================================
    # 数据库相关操作start
    # =============================================================================
    def connDB(self,config_file_path="database.conf"):  # 连接数据库
        db_host,db_user,db_pass,db_name,db_charset,db_port = self.read_config(config_file_path)
        conn = pymysql.connect(host=db_host,user=db_user,password=db_pass,db=db_name,charset=db_charset,port=int(db_port))
        cur = conn.cursor(cursor=pymysql.cursors.DictCursor)
        return conn, cur

    def exeUpdate(self,conn, cur, sql):  # 更新,插入或删除操作
        sta = cur.execute(sql)
        conn.commit()
        return sta

    def exeQuery(self,cur, sql):  # 查找操作
        cur.execute(sql)
        return cur.fetchall()


    def connClose(self,conn, cur):  # 关闭连接，释放资源
        cur.close()
        conn.close()
    # =============================================================================
    # 数据库相关操作end
    # =============================================================================

    # =============================================================================
    # 从数据库中读取要训练的数据存储到TFRecords文件中，以待模型训练使用，start    
    # =============================================================================

    def createTestFile(self,kind,date = 0):
        tableName = {0:"bidding",1:"article"}
        description = {0:"content",1:"details"}
        map_type = {0: 2, 1: 1}
        conn,cur = self.connDB()
        content_data = []
        content_id = []
        state = 1
        mark = 0
        pagesize = 100
        loop = True
        daytime = int(time.time()) - date*3600*24
        while (loop):
            sql = "select * from " + tableName[kind] + " where state = " + str(
                state) + " and FROM_UNIXTIME(time,'%Y-%m-%d') = FROM_UNIXTIME(" + str(
                daytime) + ",'%Y-%m-%d') order by id desc limit " + str(mark * pagesize) + " , " + str(pagesize)
            print(sql)
            data_cur = self.exeQuery(cur, sql)
            if data_cur.__len__() > 0:
                for data in data_cur:
                    text = data["title"] +"\t" + data[description[kind]]
                    content = self.seperate_line(self.clean_str(text))#对将要进行训练的训练集文件进行降噪处理
                    content_data.append(content)
                    content_id.append(data["id"])
            else:
                loop = False
            mark = mark + 1
            print('已经处理完第 %d 页' % (mark))

        print("待分类数据生成成功！")
        self.connClose(conn, cur)
        return content_data, content_id

    def clean_str(self, string):
        string = re.sub('\s+', "", string)
        r1 = u'[A-Za-z0-9’!"#$%&\'()*+,-./:;<=>[email protected]，。?★、…【】《》？“”‘’！[\\]^_`{|}~]+'
        string = re.sub(r1, ' ', string)
        return string.strip()

    def seperate_line(self, line):
        length = len(line)
        line = line[0:int(length / 2)]
        line = pseg.cut(line)
        new_line = []
        for words, flag in line:
            if flag == 'nr' or flag == 'ns':
                continue
            if len(flag) == 0:
                continue
            if flag[0:1] != 'n' and flag != 'v':
                # if flag[0:1] != 'n':
                continue
            new_line.append(words)
        return ''.join([word + " " for word in new_line])

    def batch_iter(self, data, batch_size, epoch_num, shuffle=True):
        data = np.array(data)
        data_size = len(data)
        batch_num_per_epoch = int((data_size - 1 / batch_size)) + 1
        for epoch in range(epoch_num):
            if shuffle:
                shuffle_indices = np.random.permutation(np.arange(data_size))
                shuffled_data = data[shuffle_indices]
            else:
                shuffled_data = data
        for batch_num in range(batch_num_per_epoch):
            start_idx = batch_num * batch_size
            end_idx = min((batch_num + 1) * batch_size, data_size)
            yield shuffled_data[start_idx: end_idx]

    def padding_sentences(self, input_sentences, padding_token, padding_sentence_length=None):
        sentences = [sentences.split() for sentences in input_sentences]
        max_sentence_length = padding_sentence_length if padding_sentence_length is not None else max(
            [len(sentence) for sentence in sentences])
        for sentence in sentences:
            if len(sentence) > max_sentence_length:
                sentence = sentence[:max_sentence_length]
            else:
                sentence.extend([padding_token] * (max_sentence_length - len(sentence)))
        return (sentences, max_sentence_length)

    def saveDict(self, input_dict, output_file):
        with open(output_file, 'wb') as f:
            pickle.dump(input_dict, f)

    def loadDict(self, dict_file):
        output_dict = None
        with open(dict_file, 'rb') as f:
            output_dict = pickle.load(f)
        return output_dict

    def check_padding_sentences(self,input_sentences, x_raw):
        valid_sentences = []
        new_x_raw = []
        valid_length = len(input_sentences[0])
        print('checking padding sentences..., valid length : ', valid_length)
        for sentence in input_sentences:
            if len(sentence) == valid_length:
                valid_sentences.append(sentence)
                new_x_raw.append(x_raw[input_sentences.index(sentence)])
        return (valid_sentences, new_x_raw)

    def createTrainFile(self,kind,filename):
        tableName = {0:"bidding",1:"article"}
        description = {0:"content",1:"details"}
        conn,cur = self.connDB()
        mark = 0
        state = 1
        pagesize = 100
        writer = tf.python_io.TFRecordWriter(os.path.join(self.base_dir, filename))
        loop = True
        while(loop):
            sql = "select * from " + tableName[kind] + " where state = "+str(state)+" limit "+str(mark*pagesize)+" , "+str( pagesize)
            data_cur = self.exeQuery(cur, sql)
            if data_cur.__len__() > 0:
                for data in data_cur:
                    text = data["title"] +"\t" + data[description[kind]]
                    reg = re.compile('<[^>]*>')
                    content = self.filter_tags(reg.sub('',text).replace('\n','').replace(' ',''))#对将要进行训练的训练集文件进行降噪处理
                    example = tf.train.Example(features=tf.train.Features(feature={
                        "label": tf.train.Feature(int64_list=tf.train.Int64List(value=[int(data["article_class_id"].replace(',',''))])),
                        'content': tf.train.Feature(bytes_list=tf.train.BytesList(value=[content.encode()]))
                    }))
                    writer.write(example.SerializeToString())  #序列化为字符串
            else:
                loop = False
            mark = mark + 1
            print('已经处理完第 %d 页' %(mark))
        writer.close()
        print("训练文件生成成功！")

    def insertPredict(self,content_id, all_pre):
        conn, cur = self.connDB()
        sqlu = "update article set article_class_id = CASE id"
        for i in range(len(content_id)):
            sqlu = sqlu + " WHEN " + str(content_id[i]) + " THEN " + str(int(all_pre[i]))
            sql = " INSERT INTO map (id, article_class_id, article_id, type, class_sort) values (0,"+str(int(all_pre[i]))+", "+str(content_id[i])+", 1, 0) "
            print(sql)
            self.exeUpdate(conn, cur, sql)
        sqlu = sqlu + " END WHERE id IN"
        ids = ''
        if len(content_id) == 1:
            ids = "(%d)" % content_id[0]
        else:
            ids = str(tuple(content_id))
        sqlu = sqlu + ids
        self.exeUpdate(conn, cur, sqlu)
        self.connClose(conn, cur)
        return 0

    # 过滤字符串中的指定字符
    def filter_tags(self, htmlstr):
        re_cdata = re.compile('//<!\[CDATA\[[^>]*//\]\]>', re.I)  # 匹配CDATA
        re_script = re.compile('<\s*script[^>]*>[^<]*<\s*/\s*script\s*>', re.I)  # Script
        re_style = re.compile('<\s*style[^>]*>[^<]*<\s*/\s*style\s*>', re.I)  # style
        re_br = re.compile('<br\s*?/?>')  # 处理换行
        re_h = re.compile('</?\w+[^>]*>')  # HTML标签
        re_comment = re.compile('<!--[^>]*-->')  # HTML注释
        blank_line = re.compile('\n+')

        # 过滤匹配内容
        s = re_cdata.sub('', htmlstr)  # 去掉CDATA
        s = re_script.sub('', s)  # 去掉SCRIPT
        s = re_style.sub('', s)  # 去掉style
        s = re_br.sub('\n', s)  # 将br转换为换行
        s = re_h.sub('', s)  # 去掉HTML 标签
        s = re_comment.sub('', s)  # 去掉HTML注释
        s = blank_line.sub('\n', s)  # 去掉多余的空行

        return s

    def is_chinese(self, uchar):
        """判断一个unicode是否是汉字"""
        if uchar >= u'\u4e00' and uchar <= u'\u9fa5':
            return True
        else:
            return False

    def format_str(self, content):
        content = np.unicode(content, 'utf-8')
        content_str = ''
        for i in content:
            if self.is_chinese(i):
                content_str = content_str + i
        return content_str           

下面对于训练文档中的数据进行细化的处理，用以提供模型训练的输入：

# encoding=utf8
import re
import jieba.posseg as pseg
import tensorflow as tf
import os
import numpy as np
import codecs
import pickle
'''
input text data and it's label
'''

tf.flags.DEFINE_integer("class_num", 2, "Number of article class to store (default: 2)")
FLAGS = tf.flags.FLAGS
FLAGS.is_parsed()
print("\nParameters:")
for attr, value in sorted(FLAGS.__flags.items()):
    print("{}={}".format(attr.upper(), value))
print("")

def load_data_and_label(file_name):
  """
  Loads MR polarity data from files, splits the data into words and generates labels.
  Returns split sentences and labels.
  """
  # Load data from files
  record_iterator = tf.python_io.tf_record_iterator(path=file_name)
  content_data = []//按照分类的个数，将文档信息存储到相应的分类数组里
  for i in range(0, FLAGS.class_num):
      content_data.append([])
  label_data = []
  labelToVec = []
  x_train = []
  classes = []
  for string_record in record_iterator:
    example = tf.train.Example()
    example.ParseFromString(string_record)
    label = example.features.feature['label'].int64_list.value[0]
    classes.append(label)
    # content_data.append(str(content.decode("utf-8")))
    content = example.features.feature['content'].bytes_list.value[0]
    content = seperate_line(clean_str(np.unicode(content, 'utf-8')))
    if (label not in labelToVec):
        labelToVec.append(label)
    content_data[labelToVec.index(label)].append(content)


  for c in content_data:
      x_train += c
  idx = 0
  for i in range(len(content_data)):
      labelvec = [0]*FLAGS.class_num
      labelvec[idx] = 1
      idx += 1
      tmplabel = [labelvec for _ in content_data[i]]
      label_data.append(tmplabel)
      # combine label
  label_data = np.concatenate(label_data, 0)
  return [x_train, label_data, labelToVec]

def clean_str(string):
    string = re.sub('\s+', "", string)
    r1 = u'[A-Za-z0-9’!"#$%&\'()*+,-./:;<=>[email protected]，。?★、…【】《》？“”‘’！[\\]^_`{|}~]+'
    string = re.sub(r1, ' ', string)
    return string.strip()

def seperate_line(line):
    length = len(line)
    line = line[0:int(length/2)]
    line = pseg.cut(line)//jieba分词
    new_line = []
    for words,flag in line:
        if flag == 'nr' or flag=='ns'://根据词性保留关键词
            continue
        if len(flag) == 0:
            continue
        if flag[0:1] !='n' and flag != 'v':
        #if flag[0:1] != 'n':
            continue
        new_line.append(words)
    return ''.join([word + " " for word in new_line])

def batch_iter(data, batch_size, epoch_num, shuffle = True):
    data = np.array(data)
    data_size = len(data)
    batch_num_per_epoch = int((data_size - 1 / batch_size)) + 1
    for epoch in range(epoch_num):
        if shuffle:
            shuffle_indices = np.random.permutation(np.arange(data_size))
            shuffled_data = data[shuffle_indices]
        else:
            shuffled_data = data
    for batch_num in range(batch_num_per_epoch):
        start_idx = batch_num * batch_size
        end_idx = min((batch_num + 1) * batch_size, data_size)
        yield shuffled_data[start_idx : end_idx]

def padding_sentences(input_sentences, padding_token, padding_sentence_length = None)://将文档统一长度，便于处理
    sentences = [sentences.split() for sentences in input_sentences]
    max_sentence_length = padding_sentence_length if padding_sentence_length is not None else max([len(sentence) for sentence in sentences])
    for sentence in sentences:
        if len(sentence) > max_sentence_length:
            sentence = sentence[:max_sentence_length]
        else:
            sentence.extend([padding_token] * (max_sentence_length - len(sentence)))
    return (sentences, max_sentence_length)

def saveDict(input_dict, output_file):
    with open(output_file, 'wb') as f:
        pickle.dump(input_dict, f)

def loadDict(dict_file):
    output_dict = None
    with open(dict_file, 'rb') as f:
        output_dict = pickle.load(f)
    return output_dict           

构建CNN神经网络：

import tensorflow as tf
import numpy as np


class TextCNN(object):
    '''
    A CNN for text classification
    Uses and embedding layer, followed by a convolutional, max-pooling and softmax layer.
    '''
    def __init__(
        self, sequence_length, num_classes,
        embedding_size, filter_sizes, num_filters, l2_reg_lambda=0.0):

        # Placeholders for input, output, dropout
        self.input_x = tf.placeholder(tf.float32, [None, sequence_length, embedding_size], name = "input_x")
        self.input_y = tf.placeholder(tf.float32, [None, num_classes], name = "input_y")
        self.dropout_keep_prob = tf.placeholder(tf.float32, name = "dropout_keep_prob")

        # Keeping track of l2 regularization loss (optional)
        l2_loss = tf.constant(0.0)

        # Embedding layer
        # self.embedded_chars = [None(batch_size), sequence_size, embedding_size]
        # self.embedded_chars = [None(batch_size), sequence_size, embedding_size, 1(num_channels)]
        self.embedded_chars = self.input_x
        self.embedded_chars_expended = tf.expand_dims(self.embedded_chars, -1)

        # Create a convolution + maxpool layer for each filter size
        pooled_outputs = []
        for i, filter_size in enumerate(filter_sizes):
            with tf.name_scope("conv-maxpool-%s" % filter_size):
                # Convolution layer
                filter_shape = [filter_size, embedding_size, 1, num_filters]
                W = tf.Variable(tf.truncated_normal(filter_shape, stddev=0.1), name="W")
                b = tf.Variable(tf.constant(0.1, shape=[num_filters]), name="b")
                conv = tf.nn.conv2d(
                    self.embedded_chars_expended,
                    W,
                    strides=[1,1,1,1],
                    padding="VALID",
                    name="conv")
                # Apply nonlinearity
                h = tf.nn.relu(tf.nn.bias_add(conv, b), name = "relu")
                # Maxpooling over the outputs
                pooled = tf.nn.max_pool(
                    h,
                    ksize=[1, sequence_length - filter_size + 1, 1, 1],
                    strides=[1,1,1,1],
                    padding="VALID",
                    name="pool")
                pooled_outputs.append(pooled)

        # Combine all the pooled features
        num_filters_total = num_filters * len(filter_sizes)
        self.h_pool = tf.concat(pooled_outputs, 3)
        self.h_pool_flat = tf.reshape(self.h_pool, [-1, num_filters_total])

        # Add dropout
        with tf.name_scope("dropout"):
            self.h_drop = tf.nn.dropout(self.h_pool_flat, self.dropout_keep_prob)

        # Final (unnomalized) scores and predictions
        with tf.name_scope("output"):
            W = tf.get_variable(
            "W",
            shape = [num_filters_total, num_classes],
            initializer = tf.contrib.layers.xavier_initializer())
            b = tf.Variable(tf.constant(0.1, shape=[num_classes], name = "b"))
            l2_loss += tf.nn.l2_loss(W)
            l2_loss += tf.nn.l2_loss(b)
            self.scores = tf.nn.xw_plus_b(self.h_drop, W, b, name = "scores")
            self.predictions = tf.argmax(self.scores, 1, name = "predictions")

        # Calculate Mean cross-entropy loss
        with tf.name_scope("loss"):
            losses = tf.nn.softmax_cross_entropy_with_logits(logits = self.scores, labels = self.input_y)
            self.loss = tf.reduce_mean(losses) + l2_reg_lambda * l2_loss

        # Accuracy
        with tf.name_scope("accuracy"):
            correct_predictions = tf.equal(self.predictions, tf.argmax(self.input_y, 1))
            self.accuracy = tf.reduce_mean(tf.cast(correct_predictions, "float"), name = "accuracy")
           

开始用该网络训练模型，并保存合适的模型以便预测使用：

#! /usr/bin/env python
# encoding: utf-8

import tensorflow as tf
import numpy as np
import os
import time
import datetime
import data_helper
import word2vec_helpers
from text_cnn import TextCNN

from scipy.sparse import csr_matrix
import random

# Parameters
# =======================================================

# Data loading parameters
tf.flags.DEFINE_float("dev_sample_percentage", .1, "Percentage of the training data to use for validation")
tf.flags.DEFINE_integer("num_labels", 2, "Number of labels for data. (default: 12)")

# Model hyperparameters
tf.flags.DEFINE_integer("embedding_dim", 128, "Dimensionality of character embedding (default: 128)")
tf.flags.DEFINE_string("filter_sizes", "3,4,5", "Comma-spearated filter sizes (default: '3,4,5')")
tf.flags.DEFINE_integer("num_filters", 128, "Number of filters per filter size (default: 128)")
tf.flags.DEFINE_float("dropout_keep_prob", 0.5, "Dropout keep probability (default: 0.5)")
tf.flags.DEFINE_float("l2_reg_lambda", 0.0, "L2 regularization lambda (default: 0.0)")

# Training paramters
tf.flags.DEFINE_integer("batch_size", 10, "Batch Size (default: 20)")
tf.flags.DEFINE_integer("num_epochs", 200, "Number of training epochs (default: 200)")
tf.flags.DEFINE_integer("evaluate_every", 100, "Evalue model on dev set after this many steps (default: 100)")
tf.flags.DEFINE_integer("checkpoint_every", 100, "Save model after this many steps (defult: 100)")
tf.flags.DEFINE_integer("num_checkpoints", 5, "Number of checkpoints to store (default: 5)")

# Misc parameters
tf.flags.DEFINE_boolean("allow_soft_placement", True, "Allow device soft device placement")
tf.flags.DEFINE_boolean("log_device_placement", False, "Log placement of ops on devices")

# Parse parameters from commands
FLAGS = tf.flags.FLAGS
FLAGS.is_parsed()
print("\nParameters:")
for attr, value in sorted(FLAGS.__flags.items()):
    print("{}={}".format(attr.upper(), value))
print("")

# Prepare output directory for models and summaries
# =======================================================

timestamp = str(int(time.time()))
out_dir = os.path.abspath(os.path.join(os.path.curdir, "runs", timestamp))
print("Writing to {}\n".format(out_dir))
if not os.path.exists(out_dir):
    os.makedirs(out_dir)

# Data preprocess
# =======================================================

# Load data
print("Loading data...")
x_text, y, labelToVec = data_helper.load_data_and_label('train.tfrecords')
print("x_text length: ", len(x_text))
print("y length: ", y.shape)

#shuffle x_text and y
#np.random.seed(10)
#x_test = np.random.permutation(x_text)
#y = np.radom.permutation(y)
#x_text = x_text[0:1000]
#y = y[0:1000]

#random select a part of the original data
new_x_text = []
new_y = []
lens = len(x_text)
for i in range(lens):
    rand_idx = random.randint(0, len(x_text)-1)
    #rand_y = random.randint(0, len(x_text))
    new_x_text.append(x_text[rand_idx])
    new_y.append(y[rand_idx])
print("new_x_text length: %d" % len(new_x_text))
print("new_y length: %d" % len(new_y))

# embedding vector
print("Padding sentences...")
sentences, max_document_length = data_helper.padding_sentences(new_x_text, '<PADDING>')    #max_document_length =

print("embedding_sentences...")
all_vectors = word2vec_helpers.embedding_sentences(sentences, embedding_size = FLAGS.embedding_dim, file_to_save = os.path.join(out_dir, 'trained_word2vec.model'))
print("all_vectors length %d  *  %d  *  %d : " %  (len(all_vectors) , len(all_vectors[0]) , len(all_vectors[0][0])))
#x = np.array(all_vectors)   ## this operation  could lead to memory error!!!

#TODO: transform large vectors into sparse matrix
x = np.asarray(all_vectors)
y = np.asarray(new_y)
print("x.shape = {}".format(x.shape))
print("y.shape = {}".format(y.shape))

# Save params
training_params_file = os.path.join(out_dir, 'training_params.pickle')
params = {'num_labels' : FLAGS.num_labels, 'max_document_length' : max_document_length, 'labelToVec' : labelToVec}
data_helper.saveDict(params, training_params_file)

# Shuffle data randomly
np.random.seed(10)
shuffle_indices = np.random.permutation(np.arange(len(y)))
x_shuffled = x[shuffle_indices]
y_shuffled = y[shuffle_indices]

# Split train/test set
# TODO: This is very crude, should use cross-validation
dev_sample_index = -1 * int(FLAGS.dev_sample_percentage * float(len(y)))
x_train, x_dev = x_shuffled[:dev_sample_index], x_shuffled[dev_sample_index:]
y_train, y_dev = y_shuffled[:dev_sample_index], y_shuffled[dev_sample_index:]
print("Train/Dev split: {:d}/{:d}".format(len(y_train), len(y_dev)))

# Training
# =======================================================
saveacc = 0
saveloss = 1
with tf.Graph().as_default():
    session_conf = tf.ConfigProto(
        allow_soft_placement = FLAGS.allow_soft_placement,
    log_device_placement = FLAGS.log_device_placement)
    sess = tf.Session(config = session_conf)
    with sess.as_default():
        cnn = TextCNN(
        sequence_length = x_train.shape[1],
        num_classes = y_train.shape[1],
        embedding_size = FLAGS.embedding_dim,
        filter_sizes = list(map(int, FLAGS.filter_sizes.split(","))),
        num_filters = FLAGS.num_filters,
        l2_reg_lambda = FLAGS.l2_reg_lambda)

    # Define Training procedure
        global_step = tf.Variable(0, name="global_step", trainable=False)
        optimizer = tf.train.AdamOptimizer(1e-3)
        grads_and_vars = optimizer.compute_gradients(cnn.loss)
        train_op = optimizer.apply_gradients(grads_and_vars, global_step=global_step)

        # Keep track of gradient values and sparsity (optional)
        grad_summaries = []
        for g, v in grads_and_vars:
            if g is not None:
                grad_hist_summary = tf.summary.histogram("{}/grad/hist".format(v.name), g)
                sparsity_summary = tf.summary.scalar("{}/grad/sparsity".format(v.name), tf.nn.zero_fraction(g))
                grad_summaries.append(grad_hist_summary)
                grad_summaries.append(sparsity_summary)
        grad_summaries_merged = tf.summary.merge(grad_summaries)

        # Output directory for models and summaries
        print("Writing to {}\n".format(out_dir))

        # Summaries for loss and accuracy
        loss_summary = tf.summary.scalar("loss", cnn.loss)
        acc_summary = tf.summary.scalar("accuracy", cnn.accuracy)

        # Train Summaries
        train_summary_op = tf.summary.merge([loss_summary, acc_summary, grad_summaries_merged])
        train_summary_dir = os.path.join(out_dir, "summaries", "train")
        train_summary_writer = tf.summary.FileWriter(train_summary_dir, sess.graph)

        # Dev summaries
        dev_summary_op = tf.summary.merge([loss_summary, acc_summary])
        dev_summary_dir = os.path.join(out_dir, "summaries", "dev")
        dev_summary_writer = tf.summary.FileWriter(dev_summary_dir, sess.graph)

        # Checkpoint directory. Tensorflow assumes this directory already exists so we need to create it
        checkpoint_dir = os.path.abspath(os.path.join(out_dir, "checkpoints"))
        checkpoint_prefix = os.path.join(checkpoint_dir, "model")
        if not os.path.exists(checkpoint_dir):
            os.makedirs(checkpoint_dir)
        saver = tf.train.Saver(tf.global_variables(), max_to_keep=FLAGS.num_checkpoints)

        # Initialize all variables
        sess.run(tf.global_variables_initializer())

        def train_step(x_batch, y_batch):
            global saveacc, saveloss
            """
            A single training step
            """
            feed_dict = {
              cnn.input_x: x_batch,
              cnn.input_y: y_batch,
              cnn.dropout_keep_prob: FLAGS.dropout_keep_prob
            }
            _, step, summaries, loss, accuracy = sess.run(
                [train_op, global_step, train_summary_op, cnn.loss, cnn.accuracy],
                feed_dict)
            time_str = datetime.datetime.now().isoformat()
            print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
            train_summary_writer.add_summary(summaries, step)
            if(saveacc < accuracy or (saveacc == accuracy and saveloss > loss)):
                saveacc = accuracy
                saveloss = loss
                path = saver.save(sess, checkpoint_prefix, global_step=step)
                print("Saved model checkpoint to {}\n".format(path))

        def dev_step(x_batch, y_batch, writer=None):
            """
            Evaluates model on a dev set
            """
            feed_dict = {
              cnn.input_x: x_batch,
              cnn.input_y: y_batch,
              cnn.dropout_keep_prob: 1.0
            }
            step, summaries, loss, accuracy = sess.run(
                [global_step, dev_summary_op, cnn.loss, cnn.accuracy],
                feed_dict)
            time_str = datetime.datetime.now().isoformat()
            print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
            if writer:
                writer.add_summary(summaries, step)

        # Generate batches
        batches = data_helper.batch_iter(
            list(zip(x_train, y_train)), FLAGS.batch_size, FLAGS.num_epochs)

        # Training loop. For each batch...
        for batch in batches:
            x_batch, y_batch = zip(*batch)
            train_step(x_batch, y_batch)
            current_step = tf.train.global_step(sess, global_step)
            if current_step % FLAGS.evaluate_every == 0:
                print("\nEvaluation:")
                dev_step(x_dev, y_dev, writer=dev_summary_writer)
                print("")
            # if current_step % FLAGS.checkpoint_every == 0:
            #     path = saver.save(sess, checkpoint_prefix, global_step=current_step)
            #     print("Saved model checkpoint to {}\n".format(path))
           

最后当然就是众望所归的预测啦~

#! /usr/bin/env python

import tensorflow as tf
import numpy as np
import os
import word2vec_helpers
import loadData as loadData
import csv

# Parameters
# ==================================================

# Eval Parameters
tf.flags.DEFINE_integer("num_labels", 12, "Number of labels for data. (default: 12)")
# Misc Parameters
tf.flags.DEFINE_boolean("allow_soft_placement", True, "Allow device soft device placement")
tf.flags.DEFINE_boolean("log_device_placement", False, "Log placement of ops on devices")
tf.flags.DEFINE_string("checkpoint_dir", ".\\article_model\\checkpoints", "Checkpoint directory from training run")
tf.flags.DEFINE_string("base_dir", ".", "files base_dir")
tf.flags.DEFINE_string("config_file", "database.conf", "database config file eg:'.conf'")
#model hyperparameters
tf.flags.DEFINE_integer('embedding_dim', 80,'dimensionality of characters')
tf.flags.DEFINE_integer("batch_size", 64, "Batch Size (default: 64)")
tf.flags.DEFINE_integer("date", 0, "date distance(default: 0)")
FLAGS = tf.flags.FLAGS
FLAGS.is_parsed()
print("\nParameters:")
for attr, value in sorted(FLAGS.__flags.items()):
    print("{}={}".format(attr.upper(), value))
print("")

# validate
# ==================================================

# validate checkout point file
checkpoint_file = tf.train.latest_checkpoint(FLAGS.checkpoint_dir)
if checkpoint_file is None:
    print("Cannot find a valid checkpoint file!")
    exit(0)
print("Using checkpoint file : {}".format(checkpoint_file))

# validate word2vec model file
trained_word2vec_model_file = os.path.join(FLAGS.checkpoint_dir, "..", "trained_word2vec.model")
if not os.path.exists(trained_word2vec_model_file):
    print("Word2vec model file \'{}\' doesn't exist!".format(trained_word2vec_model_file))
print("Using word2vec model file : {}".format(trained_word2vec_model_file))

# validate training params file
training_params_file = os.path.join(FLAGS.checkpoint_dir, "..", "training_params.pickle")
if not os.path.exists(training_params_file):
    print("Training params file \'{}\' is missing!".format(training_params_file))
print("Using training params file : {}".format(training_params_file))

# Load params
data = loadData.loadData(FLAGS.base_dir, FLAGS.config_file)
params = data.loadDict(training_params_file)
num_labels = int(params['num_labels'])
max_document_length = int(params['max_document_length'])
labelToVec = list(params['labelToVec'])
#max_document_length = 944

# Load data
x_raw, content_id = data.createTestFile(1,FLAGS.date)
# Get Embedding vector x_test
print('Padding sentence...')
sentences, max_document_length = data.padding_sentences(x_raw, '<PADDING>', padding_sentence_length = max_document_length)
print('sentences length : %d , max_document_length : %d' % (len(sentences),  max_document_length))
sentences, new_x_raw = data.check_padding_sentences(sentences, x_raw)

all_vectors = word2vec_helpers.embedding_sentences(sentences,embedding_size = 128, file_to_load = trained_word2vec_model_file)
print('all_vectors length: %d' %  len(all_vectors[0]))
x_test = np.array(all_vectors)
print("x_test.shape = {}".format(x_test.shape))
print('x_test_shape: '  , x_test.shape, " ", len(x_test) ," " , len(x_test[0]) ," " , len(x_test[0][0]))
print('list x_test ', len(list(x_test)))

# Evaluation
# ==================================================
print("\nEvaluating...\n")
checkpoint_file = tf.train.latest_checkpoint(FLAGS.checkpoint_dir)
graph = tf.Graph()
with graph.as_default():
    session_conf = tf.ConfigProto(
      allow_soft_placement=FLAGS.allow_soft_placement,
      log_device_placement=FLAGS.log_device_placement)
    sess = tf.Session(config=session_conf)
    with sess.as_default():
        # Load the saved meta graph and restore variables
        saver = tf.train.import_meta_graph("{}.meta".format(checkpoint_file))
        saver.restore(sess, checkpoint_file)

        # Get the placeholders from the graph by name
        input_x = graph.get_operation_by_name("input_x").outputs[0]
        # input_y = graph.get_operation_by_name("input_y").outputs[0]
        dropout_keep_prob = graph.get_operation_by_name("dropout_keep_prob").outputs[0]

        # Tensors we want to evaluate
        predictions = graph.get_operation_by_name("output/predictions").outputs[0]

        # Generate batches for one epoch
        batches = data.batch_iter(list(x_test), FLAGS.batch_size, 1, shuffle=False)

        # Collect the predictions here
        all_predictions = []

        for x_test_batch in batches:
            batch_predictions = sess.run(predictions, {input_x: x_test_batch, dropout_keep_prob: 1.0})
            all_predictions = np.concatenate([all_predictions, batch_predictions])

print( 'prediction data num: ', len(all_predictions))
all_pre = []
for pre in all_predictions:
    if(pre <= len(labelToVec)):
        all_pre.append(labelToVec[int(pre)])

# Save the evaluation to database
#data.insertPredict(content_id,all_pre)
print("SUCCESS!")
# Save the evaluation to a csv
predictions_human_readable = np.column_stack((np.array(new_x_raw), all_predictions, np.array(content_id)))
out_path = os.path.join(FLAGS.checkpoint_dir, "..", "prediction.csv")
print("Saving evaluation to {0}".format(out_path))
with open(out_path, 'w') as f:
    csv.writer(f).writerows(predictions_human_readable)
           

至此，整个程序就完整了，实验的结果还不错，如果有什么更好的方法，大家也可以交流一下~共同进步嘛