Pattern taxonomy mining for information filtering

Abstract

This paper examines a new approach to information filtering by using data mining method. This new model consists of two components, namely, topic filtering and pattern taxonomy mining. The aim of using topic filtering is to quickly filter out irrelevant information based on the user profiles. The aim of applying pattern taxonomy mining techniques is to rationalize the data relevance on the reduced data set. Our experiments on Reuters RCV1(Reuters Corpus Volume 1) data collection show that more effective and efficient information access has been achieved by combining the strength of information filtering and data mining method. © 2008 Springer Berlin Heidelberg.