基于无监督机器学习的抽取式文本摘要与翻译技术研究
Research on Extracted Text Summary and Translation Technology Based on Unsupervised Machine Learning
颜婷婷 1戎慧敏1
作者信息
- 1. 皖江工学院 机械工程学院,安徽 马鞍山 243000
- 折叠
摘要
翻译是促进不同语言和文化之间交流和合作的重要手段,文本摘要作为一种有效的信息提取方法,可以帮助翻译者快速准确地把握原文的核心内容和语义信息.基于此,研究引入了无监督机器学习TextRank算法应用于文本摘要抽取中,同时结合了双向编码器表示、基于相似度关系的多特征融合计算机制以及改进的最大边界相关算法加以改进.结果显示,当抽取3条摘要时,改进TextRank算法的各项Rouge值分别高达 48.01%、31.54%、37.86%.同时,改进TextRank算法在Daily-Mail数据集上双语评估研究指标高达69.81%.说明研究所提的改进TextRank算法在文本摘要抽取和翻译方面具有显著的性能优势,为现代翻译领域提供了一种有效的文本摘要抽取和翻译方法.
Abstract
Translation is an important means to promote the communication and cooperation between different languages and cultures.As an effective information extraction method,text summary can help translators to quickly and accurately grasp the core content and semantic information of the original text.Based on this,the unsupervised machine learning TextRank algorithm is applied to text summary extraction,and combines the two-way encoder representation,multi-feature fusion computer system based on similarity relationship and improved maximum boundary correlation algorithm.The results show that when three abstracts are extracted,the various Rouge values of the improved TextRank algorithm are as high as 48.01%,31.54%,and 37.86%,respectively.Meanwhile,the improved TextRank algorithm on DailyMail data set was up to 69.81%.It shows that the improved TextRank algorithm proposed has significant performance advantages in text abstract extraction and translation.It provides an effective method of text abstract extraction and translation for the modern translation field.
关键词
无监督机器学习/抽取式文本摘要/翻译技术/TextRank算法Key words
Unsupervised machine learning/Extracted text summary/Translation technology/TextRank algorithm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