News event summarization complemented by micropoints

Author

Pingping Lin ; Rong Xiao ; Yan Zhang

Author_Institution

Dept. of Machine Intell., Peking Univ., Beijing, China

fYear

2015

fDate

13-17 April 2015

Firstpage

190

Lastpage

197

Abstract

In this paper, we propose a framework to produce comprehensive summarization of news event complemented by microblog. Given a basic summary extracted from traditional news articles, we extract micropoints from microblog to complement it. A micropoint is a micro-viewpoint composed of a few sentences from several relevant microblog tweets. We first filter microblog tweets and get candidate ones according to a combination of the degree of relevance and complementarity. Then we extract micropoints through tweet clustering, scattering and micropoint rebuilding. The last step - micropoint selection is based on three criteria, which guarantee high informativeness, popularity and conciseness. Experiments conducted on real data demonstrate a good coverage and content coherence.

Keywords

Web sites; feature extraction; feature selection; information filtering; pattern clustering; relevance feedback; complementarity degree; microblog tweet filtering; micropoint extraction; micropoint rebuilding; micropoint selection; news event summarization; relevance degree; tweet clustering; tweet scattering; Clustering algorithms; Coherence; Data mining; Manuals; Mathematical model; Social network services;

fLanguage

English

Publisher

ieee

Conference_Titel

Data Engineering Workshops (ICDEW), 2015 31st IEEE International Conference on

Conference_Location

Seoul

Type

conf

DOI

10.1109/ICDEW.2015.7129575

Filename

7129575