• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>International journal of computational intelligence research >Ontology and Query-Focused Multi-Document Summarization System
【24h】

Ontology and Query-Focused Multi-Document Summarization System

机译：本体和查询为重点的多文档摘要系统

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Due to the increasing growth of online information on the specific topic, Multiple Document Summarization (MDS) has become a non-trivial task. The MDS facilitates the user to understand the large volume of information in a short time by creating a concise and comprehensive summary. In addition, user's query based MDS system provides a consistent summary, including the core of the information. The conventional summarization techniques focus on the dynamic query based summary generation. However, it lacks in providing the entire user's information in a single convenient summary according to the particular topic. As a result, it leads to the complexity of the numerous summary generation process to each query. Hence, ensuring the effective query relevant information in extractive summary is a crucial task in MDS system. To address this constraint, this paper introduces oNtology and query focused muLti documEnt summarization System (NUCLEUS). It incorporates the two essential steps such as query based summary type detection and summary generation. In the first step, NUCLEUS analyzes the document set as well as queries using ontology and Web Search Query Log (WSQL) to determine the summary type. To identify the proper context of the summary, it categorizes the document sentences based on the entities of the words in a sentence. In the second step, the NUCLEUS generates the score to each relevant query sentence using the Vector Space Model (VSM) and then, the sentences are compressed by linguistic structure analysis. Eventually, it measures the edge weight between the sentences to order coherently the sentences which having high salience and information diversity in the final summary. The evaluation results show that the NUCLEUS system can obtain significant improvement over the conventional summarization method.
机译：由于有关特定主题的在线信息的增长，多文档摘要（MDS）已成为一项艰巨的任务。 MDS通过创建简洁而全面的摘要，可帮助用户在短时间内理解大量信息。另外，基于用户查询的MDS系统提供了一致的摘要，包括信息的核心。常规的摘要技术着重于基于动态查询的摘要生成。但是，它缺乏根据特定主题在单个方便的摘要中提供整个用户信息的能力。结果，这导致针对每个查询的大量摘要生成过程的复杂性。因此，确保摘要摘要中有效的查询相关信息是MDS系统中的一项关键任务。为了解决这个限制，本文介绍了语言学和面向查询的多文档摘要系统（NUCLEUS）。它包含两个基本步骤，例如基于查询的摘要类型检测和摘要生成。第一步，NUCLEUS使用本体和Web搜索查询日志（WSQL）分析文档集以及查询，以确定摘要类型。为了确定摘要的正确上下文，它根据句子中单词的实体对文档句子进行分类。在第二步中，NUCLEUS使用向量空间模型（VSM）为每个相关查询语句生成分数，然后通过语言结构分析对这些语句进行压缩。最终，它测量句子之间的边缘权重，以在最终的摘要中一致地对具有高显着性和信息多样性的句子进行排序。评估结果表明，NUCLEUS系统与常规的摘要方法相比有明显的改进。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号