同花顺：高度口语化强背景知识的文本观点分类 baseline以及思路

赛题

高度口语化强背景知识的文本观点分类

序号	主贴内容	评论内容	分类
1	早盘收盘股价成交额在40亿左右，那么说明今天相对周五一定是放量的走势，假如收盘是涨停的话，那今天的高位震旦强分歧走一致就厉害了，明天就是缩量板的预期。且看且操作。这个位置不拉反而下杀的话达不到出货的目的，因为容易吓走套利盘，恐慌盘，止损盘，理想的剧本是边拉边出最舒服了。。。如果断板的话，也一定收红，明天看弱转强，收绿的话有点悬了。今天开盘价没有问题，高开高走也没有问题，但是后面就看分歧是否能转一致，市场合力的考验	你股龄一年，怎么这么有底气判断跌涨到什么价位？	攻击用户
2	我来梳理一下逻辑吧，说的不对大家指正，竞标应该昨晚就结束了，但是一点消息都没有，今天股价也是冲高回落，按理说主力应该比我们散户提前知道消息，中标结果一出，应该疯狂进货才对，但今天一天主力资金都是流出的，别说这个不准，参考21号龙虎榜，所以如果下午尾盘之前没有拉升迹象，就该先走为敬了	看了下主页，股龄2个月抓涨停板这么厉害！你是神仙不是人！	未攻击用户

baseline

bert一把梭
requirements

tensorflow
keras
keras-bert
#keras2.7,tensorflow2.5?按自己cuda版本来吧。

import pandas as pd
import codecs, gc
import numpy as np
from sklearn.model_selection import KFold
from keras_bert import load_trained_model_from_checkpoint, Tokenizer
# import tensorflow.keras as keras
from tensorflow.keras.metrics import top_k_categorical_accuracy,Recall
from tensorflow.keras.layers import *
from tensorflow.keras.callbacks import *
from tensorflow.keras.models import Model
import tensorflow.keras.backend as K
from tensorflow.keras.optimizers import Adam
from tensorflow.keras.utils import to_categorical

import tensorflow_addons as tfa
import tensorflow as tf
label_dict = {'未攻击用户':0,'攻击用户':1}
train_df = pd.read_table('./train/train.txt')
train_df['label'] = train_df['label'].apply(lambda x:label_dict[x])
train_df['token'] = train_df['context'] + train_df['query']

test_df = pd.read_table('./test_A/test_data_A.txt')
test_df['token'] = test_df['context'] + test_df['query']
# file_path = r'D:\ziyong\chinese_roberta_wwm_large_ext_L-24_H-1024_A-16/'
file_path = r'D:\bisai\chinese_rbtl3_L-3_H-1024_A-16/'
config_path = file_path+'bert_config_rbtl3.json'
checkpoint_path = file_path+'bert_model.ckpt'
dict_path = file_path+'vocab.txt'

maxlen=120

token_dict = {}
with codecs.open(dict_path, 'r', 'utf8') as reader:
    for line in reader:
        token = line.strip()
        token_dict[token] = len(token_dict)
 
#重写tokenizer        
class OurTokenizer(Tokenizer):
    def _tokenize(self, text):
        R = []
        for c in text:
            if c in self._token_dict:
                R.append(c)
            elif self._is_space(c):
                R.append('[unused1]')  # 用[unused1]来表示空格类字符
            else:
                R.append('[UNK]')  # 不在列表的字符用[UNK]表示
        return R
tokenizer = OurTokenizer(token_dict)
def seq_padding(X, padding=0):
    L = [len(x) for x in X]
    ML = max(L)
    return np.array([
        np.concatenate([x, [padding] * (ML - len(x))]) if len(x) < ML else x for x in X
    ])
class data_generator:
    def __init__(self, data, batch_size=32, shuffle=True):
        self.data = data
        self.batch_size = batch_size
        self.shuffle = shuffle
        self.steps = len(self.data) // self.batch_size
        if len(self.data) % self.batch_size != 0:
            self.steps += 1
 
    def __len__(self):
        return self.steps
 
    def __iter__(self):
        while True:
            idxs = list(range(len(self.data)))
 
            if self.shuffle:
                np.random.shuffle(idxs)
 
            X1, X2, Y = [], [], []
            for i in idxs:
                d = self.data[i]
                text = d[0][:maxlen]
                x1, x2 = tokenizer.encode(first=text)
                y = d[1]
                X1.append(x1)
                X2.append(x2)
                Y.append([y])
                if len(X1) == self.batch_size or i == idxs[-1]:
                    X1 = seq_padding(X1)
                    X2 = seq_padding(X2)
                    Y = seq_padding(Y)
                    yield [X1, X2], Y[:, 0, :]
                    [X1, X2, Y] = [], [], []
 
#计算top-k正确率,当预测值的前k个值中存在目标类别即认为预测正确                 
def acc_top2(y_true, y_pred):
    return top_k_categorical_accuracy(y_true, y_pred, k=2)

def micro_f1_tfaddon(y_true, y_pred, num_class=2):
    metric = tfa.metrics.F1Score(num_classes=num_class, average='micro')
    y_pred = tf.one_hot(tf.argmax(y_pred, 1), depth=num_class)
    metric.update_state(y_true, y_pred)
    result = metric.result()
    return result.numpy()
    # return result

#bert模型设置
def build_bert(nclass):
    bert_model = load_trained_model_from_checkpoint(config_path, checkpoint_path, seq_len=None)  #加载预训练模型
 
    for l in bert_model.layers:
        l.trainable = True
 
    x1_in = Input(shape=(None,))
    x2_in = Input(shape=(None,))
 
    x = bert_model([x1_in, x2_in])
    x = Lambda(lambda x: x[:, 0])(x) # 取出[CLS]对应的向量用来做分类
    x = Dense(1000,activation='relu')(x)
    p = Dense(nclass, activation='softmax')(x)
 
    model = Model([x1_in, x2_in], p)
    model.compile(loss='categorical_crossentropy',
                  optimizer=Adam(1e-5),    #用足够小的学习率
                  metrics=['accuracy', tfa.metrics.F1Score(num_classes=2, average='micro')])
    print(model.summary())
    return model
#训练数据、测试数据和标签转化为模型输入格式
DATA_LIST = []
for data_row in train_df.iloc[:].itertuples():
    DATA_LIST.append((data_row.token, to_categorical(data_row.label, 2)))
    
DATA_LIST = np.array(DATA_LIST)
 
DATA_LIST_TEST = []
for data_row in test_df.iloc[:].itertuples():
    DATA_LIST_TEST.append((data_row.token, to_categorical(0, 2)))
DATA_LIST_TEST = np.array(DATA_LIST_TEST)
 
#交叉验证训练和测试模型
def run_cv(nfold, data, data_labels, data_test):
    kf = KFold(n_splits=nfold, shuffle=True, random_state=7017).split(data)
    train_model_pred = np.zeros((len(data), 2))
    test_model_pred = np.zeros((len(data_test), 2))
 
    for i, (train_fold, test_fold) in enumerate(kf):
        X_train, X_valid, = data[train_fold, :], data[test_fold, :]
 
        model = build_bert(2)
        early_stopping = EarlyStopping(monitor='val_accuracy', patience=3)   #早停法，防止过拟合
        plateau = ReduceLROnPlateau(monitor="val_accuracy", verbose=1, mode='max', factor=0.5, patience=2) #当评价指标不在提升时，减少学习率
        checkpoint = ModelCheckpoint('./bert_dump/' + str(i)+'{epoch:02d}-{val_accuracy:.5f}-{val_f1_score:.5f}.hdf5', monitor='val_accuracy',verbose=2, save_best_only=True, mode='max', save_weights_only=True) #保存最好的模型
 
        train_D = data_generator(X_train, shuffle=True)
        valid_D = data_generator(X_valid, shuffle=True)
        test_D = data_generator(data_test, shuffle=False)
        #模型训练
        model.fit_generator(
            train_D.__iter__(),
            steps_per_epoch=len(train_D),
            epochs=5,
            validation_data=valid_D.__iter__(),
            validation_steps=len(valid_D),
            callbacks=[early_stopping, plateau, checkpoint],
        )
 
        # model.load_weights('./bert_dump/' + str(i) + '.hdf5')
 
        # return model
        train_model_pred[test_fold, :] = model.predict_generator(valid_D.__iter__(), steps=len(valid_D), verbose=1)
        test_model_pred += model.predict_generator(test_D.__iter__(), steps=len(test_D), verbose=1)
 
        del model
        gc.collect()   #清理内存
        K.clear_session()   #clear_session就是清除一个session
        # break
 
    return train_model_pred, test_model_pred
 
#n折交叉验证
train_model_pred, test_model_pred = run_cv(5, DATA_LIST, None, DATA_LIST_TEST)
test_pred = [np.argmax(x) for x in test_model_pred]

没用特征工程，而且特征比较明显，比如说带上：逼、死等脏话会对分数有提升。
可以使用树模型进行融合：https://mp.weixin.qq.com/s/S7CVNzeV8hME6qFA4sq8LA