Active reading is a process in which the reader analyses, and re-uses the object of his reading. The first step of this activity, annotating, is currently well addressed by lots of software. However, organizing annotations or building hypervideos from annotations is currently not well supported. In this paper, we describe our work about active reading of audiovisual documents. We present our analysis of active reading as an activity and we propose a model of schemas for active reading, to organize annotations and easily structure the knowledge of the reader to help him build hypervideos.