Using Topic Labels for Text Summarization
Wanqiu Kou, Fang Li
(&)
, and Zhe Ye
Department of Computer Science and Engineering,
Shanghai Jiaotong University, Shanghai 200240, People’s Republic of China
Autumn2012@qq.com, {fli,yezhejack}@sjtu.edu.cn
Abstract. Multi-document summarization is a difficult natural language pro-
cessing task. Many extractive summarization methods consist of two steps:
extract important concepts of documents and select sentences based on those
concepts. In this paper, we introduce a method to use the Latent Dirichlet
Allocation (LDA) topic labels as concepts, instead of n-gram or using external
resources. Sentences are selected based on these topic labels in order to form a
summary. Two selection methods are proposed in the paper. Experiments on
DUC2004 dataset has shown that Vector-based methods are better, i.e. map
topic labels and sentences to a word vector and a letter trigram vector space to
find those sentences which are syntactically and semantically related with the
topic labels in order to form a summary. Experiments show that the produced
summaries are informative, abstractive and better than the baseline method.
Keywords: Text summarization
Topic labels Word vectors
1 Introduction
With the rapid development of the Internet, information has witnessed explosive
growth. To browse and search information in an effective way has becom e an important
issue in natural language processing. Automatic text summarization can compress
document information and help users absorb mass information. Such technologies can
decrease information overload effectively.
Extractive and abstractive ways are usually two kinds methods for automatic
summarization (Ani Nenkova and Kathleen McKeown 2011). Considering the diffi-
culty of abstractive way, most researchers use many different extractive methods to
extract some important sentences as text summarization, such as supervised method (Li
et al. 2013), graph based method (Erkan and Dragomir 2004), global optimization
method (Dimitrios et al. 2012) and concept based method (Gillick and Favre 2009).
It has been assumed that the value of a summary is the sum of the values of the
unique concepts it contains. Concepts could be words, named entities, syntactic sub-
trees or semantic relations. The goal is to maximize the sum of the weights of those
concepts that will be chosen to appear in the summary (Gillick and Favre 2009).
In this paper, we propose a method for concept based multi-document text summa-
rization. LDA topic labels are used as concepts. Our method consists of three steps: use
LDA topic model to generate LDA topics, then generate topic labels by vector based
method, select sentences that are semantically related with topic labels as final summaries.
© Springer International Publishing AG 2017
S. Benferhat et al. (Eds.): IEA/AIE 2017, Part II, LNAI 10351, pp. 448–457, 2017.
DOI: 10.1007/978-3-319-60045-1_46
下载后可阅读完整内容，剩余9页未读，立即下载
weixin_38702515
- 粉丝: 12
- 资源: 927
我的内容管理
展开
最新资源
- OptiX传输试题与SDH基础知识
- C++Builder函数详解与应用
- Linux shell (bash) 文件与字符串比较运算符详解
- Adam Gawne-Cain解读英文版WKT格式与常见投影标准
- dos命令详解：基础操作与网络测试必备
- Windows 蓝屏代码解析与处理指南
- PSoC CY8C24533在电动自行车控制器设计中的应用
- PHP整合FCKeditor网页编辑器教程
- Java Swing计算器源码示例：初学者入门教程
- Eclipse平台上的可视化开发：使用VEP与SWT
- 软件工程CASE工具实践指南
- AIX LVM详解：网络存储架构与管理
- 递归算法解析：文件系统、XML与树图
- 使用Struts2与MySQL构建Web登录验证教程
- PHP5 CLI模式：用PHP编写Shell脚本教程
- MyBatis与Spring完美整合：1.0.0-RC3详解
资源上传下载、课程学习等过程中有任何疑问或建议，欢迎提出宝贵意见哦~我们会及时处理！
点击此处反馈
