A user-oriented splog filtering based on a machine learning

  • Authors:

  • Takayuki Yoshinaka;Soichi Ishii;Tomohiro Fukuhara;Hidetaka Masuda;Hiroshi Nakagawa

  • Affiliations:

  • School of Science and Technology for Future Life, Tokyo Denki University, Tokyo, Japan;School of Science and Technology for Future Life, Tokyo Denki University, Tokyo, Japan;Research into Artifacts, Center for Engineering, The University of Tokyo, Kashiwa, Chiba, Japan;School of Science and Technology for Future Life, Tokyo Denki University, Tokyo, Japan;Information Technology Center, The University of Tokyo, Tokyo, Japan

  • Venue:
  • BlogTalk'08/09 Proceedings of the 2008/2009 international conference on Social software: recent trends and developments in social software
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

A method for filtering spam blogs (splogs) based on a machine learning technique, and its evaluation results are described. Today, spam blogs (splogs) became one of major issues on theWeb. The problem of splogs is that values of blog sites are different by people. We propose a novel user-oriented splog filtering method that can adapt each user's preference for valuable blogs. We use the SVM(Support Vector Machine) for creating a personalized splog filter for each user. We had two experiments: (1) an experiment of individual splog judgement, and (2) an experiment for user oriented splog filtering. From the former experiment, we found existence of 'gray' blogs that are needed to treat by persons. From the latter experiment, we found that we can provide appropriate personalized filters by choosing the best feature set for each user. An overview of proposed method, and evaluation results are described.