近70年文本自动摘要研究综述

发布时间：2018-02-17 05:26

本文关键词： 自动摘要机器学习语言网络深度学习综述　出处：《情报科学》2017年07期 　论文类型：期刊论文

【摘要】：【目的/意义】文本自动摘要能快速获取文本主要内容,极大提高信息使用效率,帮助人们从信息海洋中解放出来。随着互联网大数据日益深入发展,文本信息的数量已经远远超出人工处理极限,文本自动摘要研究显得越发迫切和重要。【方法/内容】本文通过对过去70年国内外文本自动摘要经典文献重要文献进行收集、整理和分析,总结归纳出六类主要文本自动摘要方法及其理念和具体做法,对比评析其优势不足,并对未来研究方向进行展望,绘制出该研究领域的一个发展全景图。【结果/结论】自动摘要方法所使用的特征经历了由简单到复杂、由个体到联系、由表层到深层的发展路径;如何深入利用深度学习方法,以及如何将传统方法与深度学习方法更好结合起来将是下一步研究的热点和重点。
[Abstract]:[purpose / meaning] automatic text abstracts can quickly obtain the main contents of the text, greatly improve the efficiency of the use of information, and help people to free themselves from the ocean of information. With the deepening development of big data on the Internet, The amount of text information has gone far beyond the limit of manual processing, and the study of automatic text abstracts has become more and more urgent and important. [methods / contents] this paper collects the classical literature on automatic text abstracts from home and abroad in the past 70 years. In this paper, the author summarizes six kinds of automatic text summarization methods, their concepts and concrete practices, compares their advantages and disadvantages, and looks forward to the future research direction. Drawing a panorama of developments in the field of research. [results / conclusions] the features used in the automated summary method have experienced a path from simplicity to complexity, from individual to connection, from surface to deep; how to make use of the in-depth learning method, And how to better combine the traditional method with the deep learning method will be the focus of the next research.
【作者单位】：中国科学院文献情报中心;中国科学院大学;浙江师范大学经济与管理学院;
【基金】：教育部人文社科基金项目(14YJC870029) 国家社会科学基金项目(15CTQ022)
【分类号】：G254.3;TP181

【参考文献】