摘要
随着新型冠状病毒在全球的爆发,人们越来越重视生命健康与用药安全。近些年,生物医学科研领域呈现快速发展趋势,形成了庞大的文本数据资源。本文聚焦于中文医学文本数据,提出了Bert-wwm编码与Casrel标注器解码相结合的抽取模型,并与原模型进行了实验对比。结果表明,基于该模型的F1值为63.3%,与基础模型相比有了进一步的提升。
With the global outbreak of novel coronavirus,people pay more and more attention to life health and medication safety.In recent years,with the publication of a large number of valuable literature that forms a huge text data resource,the biomedical research field has also shown a rapid development trend.This article focuses on Chinese medical text data and proposes a extraction model that combines Bert-wwm-ext encoder and Casrel tagging decoder,making an experimental comparison with the original model.The results show that the F1 value based on this model is 63.3%,which is further improved compared with the original one.
作者
姜智尹
程翔
JIANG Zhiyin;CHENG Xiang(School of Information Engineering,Jingdezhen Ceramic University,Jingdezhen,China,333403)
出处
《福建电脑》
2023年第8期56-58,共3页
Journal of Fujian Computer
关键词
中文医学文本挖掘
关系抽取
预训练模型
层级标注器
Chinese Medical Text Mining
Relation Extraction
Pretraining Model
Cascade Novel Tagger
