Affordable Access


  • Nicolas Sauret


ACM short paper : VidéoLivre 1 ACM 2010 : short paper VidéoLivre / Book-Video collaborative annotation, polemical semantic and dynamic ontologies Video Corpus The concept of Video-Livre or Book-Video was motivated by the need of an indexed corpus of video-interviews. Indexed content opens the possibility to access and to manipulate the content easily and thus to open it to critics and intellectual discourse. Therefore the video content would eventually obtain a status of academical corpus in the same way the printed corpus. Video Template The original plan was to rationalize the format of the interviews in order to allow pertinent indexation. We focused first on trying to standardize the interviews through a template of questions in order to make an automatic pre-segmentation and tag-indexation possible. The template carries editorial decisions that supposedly structure the content and set a top-down control on the content. This top-down feature characterizing at first the Video-Livre is an attempt to balance pure bottom-up content, which is often considered as lacking long term impact because of its inability to exist within categorized corpus. It is an attempt to higher the level of personal contribution. Pre-segmentation would be the first step to video-chapters with a common base of tag for each chapter. The Video-Livre would be then open to a a more interpretative level of annotations through the online platform of consultation, annotation and eventually auto-production. However the constraint of a template applied to speech appeared rapidly to be inappropriate to direct expression to a camera. Pre-segmentation and indexation remains an important aspect for a balanced top-down and bottom-up articulation but the core of the problematic seems to take place at a higher level of abstraction such as the metadata format rather than in the Video-Livre format or template. Collaborative Annotation In any case, remained the necessity to organize the debate and the intellectual discourse around the video corpus. The issue pushed simultaneously our team to bring deep-annotation tools such as Ligne de Temps towards a more collaborative approach with shared annotations and contribution through an online platform of consultation and annotation. Lignes de Temps is a standalone video annotation tool based on a format of metadatas specific to video or " temporal object " content. Video-Livre inherits from the innovations brought by Ligne de Temps and integrates visually the metadatas production and consultation within the content itself.

Report this publication


Seen <100 times