基于滑动窗口的微博时间线摘要算法

发布时间：2018-03-22 06:00

本文选题：微博摘要　切入点：时间线摘要　出处：《数据采集与处理》2017年03期 　论文类型：期刊论文

【摘要】：是在时间维度上对文本进行内容归纳和概要生成的技术。传统的时间线摘要主要研究诸如新闻之类的长文本,而本文研究微博短文本的时间线摘要问题。由于微博短文本内容特征有限,无法仅依靠文本内容生成摘要,本文采用内容覆盖性、时间分布性和传播影响力3种指标评价时间线摘要,并提出了基于滑动窗口的微博时间线摘要算法(Microblog timeline summariaztion based on sliding window,MTSW)。该算法首先利用词项强度和熵来确定代表性词项;然后基于上述3种指标构建出评价时间线摘要的综合评价指标;最后采用滑动窗口的方法,遍历时间轴上的微博消息序列,生成微博时间线摘要。利用真实微博数据集的实验结果表明,MTSW算法生成的时间线摘要可以有效地反映热点事件发展演化的过程。
[Abstract]:Traditional time-line abstracts focus on long texts such as news. In this paper, we study the time line summary of Weibo short text. Due to the limited content features of the short text, we can not only rely on the text content to generate the summary. Time distribution and spreading influence are used to evaluate time-line abstracts, and Weibo time-line summarization algorithm based on sliding window is proposed. Firstly, the term intensity and entropy are used to determine representative word items. Then, based on the above three indexes, a comprehensive evaluation index of the summary of the evaluation time line is constructed. Finally, the sliding window is used to traverse the Weibo message sequence on the time axis. The experimental results show that the time line summary generated by the MTSW algorithm can effectively reflect the development and evolution of hot events.
【作者单位】：南京师范大学计算机科学与技术学院;
【基金】：江苏省高校自然科学基金(13KJB520014)资助项目
【分类号】：TP391.1;TP393.092

【相似文献】