NLP15：使用BiLSTM、BiLSTM-CRF、BiLSTM-Attention、Bert-BiLSTM-CRF进行命名实体识别

最新推荐文章于 2024-07-31 10:33:30 发布

大雄没有叮当猫

最新推荐文章于 2024-07-31 10:33:30 发布

阅读量6.9k

点赞数 13

分类专栏：自然语言处理命名实体识别

本文链接：https://blog.csdn.net/u013230189/article/details/110007826

版权

公众号：数据挖掘与机器学习笔记

1.基于BiLSTM的命名实体识别

Embedding+BiLSTM+BiLSTM+Dense

from tensorflow.keras.layers import Embedding, LSTM, Bidirectional, Dense, Dropout, Masking
from tensorflow.keras.models import Sequential


def build_model():
    """
    使用Sequential构建模型网络:双向LSTM
    :return:
    """
    model = Sequential()
    model.add(Embedding(input_dim=vocab_size, output_dim=embedding_dim))
    model.add(Masking(mask_value=0))
    model.add(Bidirectional(LSTM(lstm_hidden_size, return_sequences=True)))
    model.add(Dropout(dropout_rate))
    model.add(Bidirectional(LSTM(lstm_hidden_size, return_sequences=True)))
    model.add(Dropout(dropout_rate))
    model.add(Dense(tag_size, activation="softmax"))
    model.compile(loss=loss, optimizer=optimizer, metrics=[metrics])
    model.summary()
    return model


if __name__ == '__main__':
    model = build_model()
    train_x = np.array(padding_sequences_ids)
    train_y = np.array(padding_tags_ids)
    # mod