Named entity identification method for Chinese medical record of iterative expansion convolutional neural network-conditional random field based on word structure

Abstract

The invention discloses a named entity recognition method for Chinese medical records of an iterative expansion convolutional neural network-conditional random field based on a word structure. The named entity recognition method comprises the following steps: 1) extracting feature vectors from bitmaps corresponding to Chinese characters through a convolutional neural network for a training data set of a group of inquiry sequences and entity labeling sequences; 2) combining a word embedding result with a feature vector output by the convolutional neural network; 3) obtaining a score sequence for each label in the label set through an iterative expansion convolutional neural network and an attention mechanism; and 4) obtaining a named entity identification result through a linear chain conditional random field algorithm. According to the named entity identification method, the Chinese character structure information in the Chinese medical record can be utilized, and the iterative expansion convolutional neural network and the conditional random field algorithm are combined, so that the performance of the named entity identification method can be further improved.

Description

technical field [0001] The invention relates to the field of natural language processing named entity recognition, in particular to a named entity recognition method for Chinese medical records based on word structure-based iterative expansion convolutional neural network-conditional random field. Background technique [0002] Medical records are very important research data, but manual analysis of medical records is time-consuming and expensive. Therefore, automatic and efficient machine learning algorithms are critical applications in this field. Named entity recognition is to find entities with special meaning under the given content, and it is the basis of analytical work such as relation extraction. [0003] Currently, the most advanced named entity recognition models include BiLSTM-CNN-CRF proposed by Ma and Hovy in 2016, and IDCNN-CNN-CRF proposed by Strubell et al. in 2017. The above two models combine word embeddings and word-level features to improve performance....

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a Chinese medical record named entity recognition method based on word structure-based iterative expansion convolutional neural network-conditional random field, which solves the problem of Chinese character structure information loss in the word embedding process, and improves the efficiency of electronic medical treatment. Recorded Named Entity Recognition Performance

Images