首页> 外国专利> NAMED ENTITY RECOGNITION METHOD AND APPARATUS, TERMINAL DEVICE AND STORAGE MEDIUM

NAMED ENTITY RECOGNITION METHOD AND APPARATUS, TERMINAL DEVICE AND STORAGE MEDIUM

机译:命名实体识别方法和装置,终端设备和存储介质

摘要

A named entity recognition method and apparatus, a terminal device and a storage medium, which are applicable to the technical field of computers. Said method comprises: acquiring text to be recognized, and converting said text into a n×k-dimensional first matrix (S301); performing convolution of multiple convolution layers on the first matrix, wherein a convolution operation is performed four times on the last convolution layer, of which the number of channels of the convolution kernel is m, so as to obtain four parallel n×m-dimensional second matrices (S302); performing attention weight adaptation on three second matrices of the four second matrices to obtain an n×m-dimensional third matrix, and performing matrix addition on the third matrix and the remaining second matrix to output an n×m-dimensional fourth matrix (S303); classifying the fourth matrix, and outputting an entity label corresponding to said text; and according to the entity label, outputting a named entity corresponding to said text. By introducing an attention mechanism in the convolution layer, the data redundancy is effectively reduced, and the recognition speed of named entities is increased.
机译:一种名为实体识别方法和装置,终端设备和存储介质,其适用于计算机技术领域。所述方法包括:获取要识别的文本,并将所述文本转换为N×K维第一矩阵(S301);在第一矩阵上执行多个卷积层的卷积,其中在最后一个卷积层上执行卷积操作,其中卷积内核的信道的数量是m,以便获得四个并行n×m维秒矩阵(S302);在四个第二矩阵的三个第二矩阵上执行注意力适应,以获得N×M维第三矩阵,并执行第三矩阵的矩阵和剩余的第二矩阵输出N×M维第四矩阵(S303) ;分类第四矩阵,并输出与所述文本相对应的实体标签;根据实体标签,输出与所述文本相对应的命名实体。通过在卷积层中引入注意机制,有效地减少了数据冗余,并增加了命名实体的识别速度。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号