Skip to main navigation Skip to search Skip to main content

Chinese spam filter based on relaxed online support vector machine

  • Yong Han*
  • , Xiaoning He
  • , Muyun Yang
  • , Haoliang Qi
  • , Chao Song
  • *Corresponding author for this work
  • Heilongjiang Institute of Technology
  • School of Computer Science and Technology, Harbin Institute of Technology

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

Abstract

For spam filtering, a classical online learning problem, Online SVM is less applicable for its low efficiency in dealing with the ever increasing training sample set. For this purpose, the Relaxed Online SVM (ROSVM) model by relaxing its constraints can significantly improve training speed with little cost on filtering performance. In this paper, we applied this model to Chinese spam filtering and experimental results have outperformed the best in the TREC 2006 Chinese spam filtering track. Our filter is further validated by the SEWM 2010 dataset, ranking the best according to 1-ROCA% in the delayed feedback task and the active learning task.

Original languageEnglish
Title of host publicationProceedings - 2010 International Conference on Asian Language Processing, IALP 2010
Pages185-188
Number of pages4
DOIs
StatePublished - 2010
Externally publishedYes
Event2010 International Conference on Asian Language Processing, IALP 2010 - Harbin, China
Duration: 28 Dec 201030 Dec 2010

Publication series

NameProceedings - 2010 International Conference on Asian Language Processing, IALP 2010

Conference

Conference2010 International Conference on Asian Language Processing, IALP 2010
Country/TerritoryChina
CityHarbin
Period28/12/1030/12/10

Keywords

  • Chinese spam filtering
  • Online learning
  • Relaxed online SVM

Fingerprint

Dive into the research topics of 'Chinese spam filter based on relaxed online support vector machine'. Together they form a unique fingerprint.

Cite this