基于LATTICE的文言文自动断句研究
Research on Automatic Ancient Chinese Texts Segmentation Based on LATTICE
庄百川 1于文年 2邱秀连2
作者信息
- 1. 武汉邮电科学研究院 武汉 430074;南京烽火天地通信科技有限公司 南京 210019
- 2. 南京烽火天地通信科技有限公司 南京 210019
- 折叠
摘要
随着社会主义文化强国的建设,利用自然语言处理技术实现对古汉语典籍的处理与挖掘越发受到关注.针对文言文自动化断句任务,论文提出一种基于BERT-LATTICE-LAN网络的深度学习模型,通过集成预训练模型和融入词向量的长短时记忆网络,实现对文言文语句的自动化断句处理.相比于传统的BiLSTM-CRF网络,其准确率和F值分别提高了8.03%和7.88%.
Abstract
With the construction of socialist cultural power,the use of natural language processing technology to realize the processing and excavation of ancient Chinese classics has attracted more and more attention.In view of the task of automating sen-tence breaks in text,this paper proposes a deep learning model based on BERT-LATTICE-LAN network,which can realize the au-tomated break processing of text statements by integrating the pretrained model and the long-term memory network integrated into the word vector.Compared to traditional BiLSTM-CRF networks,the accuracy and F values are increased by 8.03 percent and 7.88 percent respectively.
关键词
深度学习/BERT/LATTICE/序列标注/自动断句Key words
deep learning/BERT/LATTICE/sequence tagging/automatic segmentation引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据