首页> 外文会议>IFAC World Congress >Topic analysis in news via sparse learning: a case study on the 2016 US presidential elections
【24h】

Topic analysis in news via sparse learning: a case study on the 2016 US presidential elections

机译:通过稀疏学习的新闻主题分析 - 以2016年美国总统大选为例

获取原文

摘要

Textual data such as tweets and news is abundant on the web. However, extracting useful information from such a deluge of data is hardly possible for a human. In this paper, we discuss automated text analysis methods based on sparse optimization. In particular, we use sparse PCA and Elastic Net regression for extracting intelligible topics from a big textual corpus and for obtaining time-based signals quantifying the strength of each topic in time. These signals can then be used as regressors for modeling or predicting other related numerical indices. We applied this setup to the analysis of the topics that arose during the 2016 US presidential elections, and we used the topic strength signals in order to model their influence on the election polls.
机译:Web上的文本数据如推文和新闻。然而,对于人类来说,从这种酝酿数据中提取有用的信息。在本文中,我们讨论了基于稀疏优化的自动化文本分析方法。特别是,我们使用稀疏的PCA和弹性净回归来从大型文本语料库中提取可理解的主题,并以获取时间的基于时间的信号量化每个主题的强度。然后,这些信号可以用作用于建模或预测其他相关数字索引的回归器。我们将此设置应用于分析2016年美国总统选举期间出现的主题,我们使用了主题强度信号,以便为选举民意调查进行影响。

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