Sentence Level scenario detector_CNN #CNN #NLP

sentence_level_detector_CNN.py

import os

import numpy as np
import pandas as pd
from bert_serving.client import BertClient
from keras.layers import Conv1D
from keras.layers import Dense, Activation, Dropout, Flatten, AveragePooling1D
from keras.models import Sequential
from keras.optimizers import RMSprop
from keras.utils import np_utils

np.random.seed(1337)
bc = BertClient(check_version=False)
sentence_path = os.path.join("/Users/elfsong/PycharmProjects/BERT_demo", "sentence_class.xls")
data_frame = pd.read_excel(sentence_path, sheet_name='sheet1')

train_data_list = list()
train_lable_list = list(data_frame["type"])

for sentence in data_frame["sentence"]:
    print(sentence)
    result = bc.encode([sentence])
    train_data_list += [result[0]]

# 数据预处理
X_train = np.array(train_data_list)
X_train = np.expand_dims(X_train, 2)
y_train = np_utils.to_categorical(train_lable_list, num_classes=11)
print(X_train.shape)
print(y_train.shape)

# 模型构建
model = Sequential([
    Conv1D(filters=5, kernel_size=5, strides=1, padding='valid', input_shape=(768, 1), name="Convolution_Layer_1"),
    AveragePooling1D(pool_size=5, strides=1, padding="valid", name="Pooling_Layer_1"),

    Conv1D(filters=5, kernel_size=5, strides=1, padding='valid', name="Convolution_Layer_2"),
    AveragePooling1D(pool_size=5, strides=1, padding="valid", name="Pooling_Layer_2"),

    Flatten(name="Flatten_Layer"),

    Dense(256, input_dim=3760, name="Dense_Layer_1"),
    Activation('relu'),
    Dropout(0.1),

    Dense(32, input_dim=256, name="Dense_Layer_2"),
    Activation('relu'),
    Dropout(0.1),

    Dense(11, input_dim=32, name="Dense_Layer_3"),
    Activation('softmax'),
])

# 激活RMS优化器
rmsprop = RMSprop(lr=0.001, rho=0.9, epsilon=1e-08, decay=0.0)
model.compile(optimizer=rmsprop,
              loss='categorical_crossentropy',
              metrics=['accuracy'])

model.fit(X_train, y_train, epochs=120, batch_size=8)


def get_result(result):
    max_index = np.argmax(result)
    category_list = ["None", "city", "forest", "default", "flatland", "river", "college", "town", "mountain", "ocean",
                     "plaza"]
    return category_list[max_index]


sentence_vector = bc.encode(["猫爷爷带着[小鸭子]来到了动物游乐场。"])
sentence_vector = np.expand_dims(sentence_vector, 2)
result = model.predict(sentence_vector)

print(get_result(result))