Personalized peer filtering for a dynamic information push

Abstract

Due to the anonymity of the user during Web searching, no support for long-term information needs exists. First attempts for personalized Web retrieval are made, however these approaches are limited to static objects and no individual recommendations from a dynamic data set can be determined. Peer-to-peer architectures build a promising platform for a personalized information filtering system, where all steps during information exchange are transparent to the user. Our approach assists active requests in the form of an information pull as well as a system initiated information push. In a cooperative manner all peers have the function of information providers and consumers. The ranking of recommendations is established by a community-based filtering approach. © Springer-Verlag Berlin Heidelberg 2005.