In this paper, we present work in cross-media summarization which is currently in progress under the REVEAL-THIS IST project. The proposed strategy emphasizes on exploring different ways of synthesizing the most salient elements of the constituent parts of a cross-media object. At the core of our work lies an open, adaptable architecture that decides the way the salient parts are fused in accordance with both the users’ interests and digital equipment and the typology and semantic characteristics of the original information. We apply our methodology by showing specific examples of our notion of cross-media fusion and summarization in the domains of broadcast TV news and European Parliament sessions.