[Ce séminaire est une "pratique" pour une présentation qui sera donnée au TAC2010 Workshop le 15 novembre] I will describe Rali's participation at this year's competition. Our system extracts Information Items from the sentences in the source documents (rather than extracting full sentences) and generates short, to-the-point sentences from them. Information Extraction was used to deal directly with locations and dates, and some text planning has been performed to handle this. I will also discuss in some detail our longer-term goals for the development of a fully abstractive summarization system.