Skip to main navigation Skip to search Skip to main content

Event detection and recommendation based on heterogeneous information

  • Bo Yuan*
  • , Qingcai Chen
  • , Yang Xiang
  • , Xiaolong Wang
  • *Corresponding author for this work
  • Harbin Institute of Technology Shenzhen

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Previous research on event detection only handles with text data, and there is still no agreed standard except for the reports amount on judging whether a set of information should be pushed as a hot event. In this paper, we present an event detection framework based on heterogeneous information. Firstly, the coarse classification of structured data is transplanted to text data to make the information set more precise, and then twice clustering using multi-features is attempted to enhance the performance of event detection. Meanwhile, data fluctuation of structured data is monitored to determine the event priority. The experiment results and online system proved the availability of our method.

Original languageEnglish
Title of host publicationProceedings of the International Conference on Information Engineering and Applications, IEA 2012
Pages407-416
Number of pages10
EditionVOL. 2
DOIs
StatePublished - 2013
Externally publishedYes
Event2nd International Conference on Information Engineering and Applications, IEA 2012 - Chongqing, China
Duration: 26 Oct 201228 Oct 2012

Publication series

NameLecture Notes in Electrical Engineering
NumberVOL. 2
Volume217 LNEE
ISSN (Print)1876-1100
ISSN (Electronic)1876-1119

Conference

Conference2nd International Conference on Information Engineering and Applications, IEA 2012
Country/TerritoryChina
CityChongqing
Period26/10/1228/10/12

Keywords

  • Clustering
  • Event detection
  • Heterogeneous information
  • Multi-features

Fingerprint

Dive into the research topics of 'Event detection and recommendation based on heterogeneous information'. Together they form a unique fingerprint.

Cite this