Evolving information filtering for personalized information service

Abstract

Information filtering (IF) systems are important for personalized information service. However, most current IF systems suffer from low quality and long training time. In this paper, a refined evolving information filtering method is presented. This method describes user’s information need from multi-aspects and improves filtering quality through a process like natural selection. Experimental result shows this method can shorten training time, improve filtering quality, and reduce the relevance between filtering results and training sequence.