Vision of the crowds

Vision of the Crowds is a paraphrase of the well-known notion Wisdom of the Crowds. In the latter, instead of using an ontology-based, top-down approach, we create knowledge rather bottom-up, relying on the knowledge of the members of a certain community. Well-known applications of this approach are cooperative tagging of photos or filtering spams. In the case of Vision of the Crowds, we do not insist on creating new knowledge; a new, cooperative view may also suffice for many purposes. Take as an example an open-air sport event, which happens on different locations over several hours or days. Many thousands of visitors make audio-visual recordings. Nobody has a full view of the event, but we have a large amount of partial views. The challenge is to provide concepts and tools with the help of which such partial views can be shared and automatically summarized corresponding to different intentions of the viewers. The talk describes a set of ideas and presents some first results of realizing the idea of the Vision of the Crowds. Beside general considerations, two particular applications will be introduced, in the area of medicine (in particular endoscopic videos) and traffic.