Tag based generation of user profiles

This article investigates the use of tags for the automatic generation of user profiles. An architecture has been created to collect these tags from different social websites, process them and place them into hierarchical categories. This is achieved by using a combination of preprocessing steps, clustering techniques and consultation of an online taxonomy. We have tested this approach on tagsets from four very popular web 2.0 sites. Results for Delicious are promising, but for Flickr, Last.fm and YouTube results are less good due to the nature of their folksonomies and the kind of content they contain