Chinese Medical Named Entity Recognition Based on Chinese Character Radical Features and Pre-trained Language Models

Abstract

With the continuous innovation of medical diagnosis forms, the communication and interaction between patients and medical staff have become more frequent and informationalized. How to extract valuable information from a large amount of clinical medical text has attracted more and more attention of researchers. Medical named entity recognition is the basis for structuring the information in clinical medical text, and has great technical significance. This paper aims at the task of medical named entity recognition, using Bidirectional Long Short-Term Memory, conditional random field, adapting Transformer encoder for named entity recognition, incorporating Chinese character radical information, and converting it into machine reading comprehension task, to build medical named entity recognition models based on different pre-trained language models. Finally, the model ensemble is carried out on the results of the above multiple different models to obtain the final recognition result. In the final test data set of medical term recognition track of “Chinese Clinical Medical Text Word Segmentation and Named Entity Challenge”, the method in this paper obtains the F1 score of 82.72%.