Web 2.0 applications such as Flickr offer a rich set of data with a huge potential for exploitation by the human users. Unfortunately, the sifting through such data is far from easy and rewarding due to a lack of semantics on the one side and a lack of rich data description on the other side. For instance, most photos on Flickr have very little description attached that could be used for retrieving or exploring the photos. In this talk, we demonstrate how the enrichment of Web 2.0 data by automatically discovered (more or less) semantic relationships improves the user experience.