期刊论文

中文

计算机工程

1000-3428

2018

44

7

172-176

10.19678/j.issn.1000-3428.0047156

国家自然科学基金青年基金（61402191） 国家语委科研项目（WT135-11） 华中师范大学中央高校基本科研业务费专项资金（CCNU16JYKX15）

本校为第一机构

时间轴摘要可帮助用户获取感兴趣的新闻话题发展轨迹,但现有研究中大多仅考虑句子间的关系来对句子进行打分排序,忽视了文档主题层面的信息影响。为此,提出一种新的基于子主题增强的摘要算法。考虑句子间的关系,分析每个时间段内的子主题对句子的影响,使得与重要子主题越相关的句子得分越高,通过句子与子主题的互强化来对句子进行主题层面的综合打分排序。实验结果表明,与现有的时间轴摘要算法相比,该算法可移植性较好,且可准确获取新闻演化轨迹。

The Timeline Summary ( TS) can help users obtain the development trajectory of interesting news topics. However,most of the existing researches only consider the relationship between sentences to score sentences and ignore the information impact on the topic level of documents. For this reason,a new subtopic-based summarization method is proposed. Considering the relationship between sentences and introducing the influence of subtopics in each time period on the sentences, and to make the score of the sentences higher which is more related to the important subtopic,it sorts the sentence compre...