摘要
【目的/意义】多文档自动摘要技术的目的是从一组文档中精炼出重要信息摘要,减轻用户从文档中获取与理解信息的负担,是自然语言理解领域的重要研究方向之一。【方法/过程】本文提取十五年内的多文档自动摘要研究文献并筛选出至少50篇关键影响文章,梳理多文档自动摘要的概念与研究进展,揭示了最新的技术实现与实践情况。【结果/结论】基于不同技术方法对单词、句子或段落作为主要数据处理对象,找出多文档自动摘要的技术特征与难点,明确该领域的发展趋势,为未来的研究奠定了基础。
[Purpose/significance]Multi-document automatic summarization (MDS) extracts the most essential information from a group of documents and provides users to understand and exploit information easily and quickly. MDS has been one of the most important studies of natural language understanding recently. [Purpose/significance] More than 50 papers selected from key journals and conferences between 2000 and 2015 are reviewed, and the concepts, state-of-the-art methodologies, improvements of MDS are displayed and concluded in this article. [ Result/conclusion ] Features and difficulties of MDS methods are separated and analyzed in word-level, sentence-level, and paragraph-level to show its trend for further researches.
出处
《情报科学》
CSSCI
北大核心
2017年第4期160-165,共6页
Information Science
基金
中国科技信息研究所项目合作支持
关键词
自动摘要
多文档处理
自然语言处理
document automatic summarization
multi-document processing
natural language processing