Glocal Multimedia Retrieval

Personal experience is intrinsically local while common knowledge is global. As a consequence, standard multimedia search engines suffer from a gap between local content and global concept, due to the diversity of context. Here we design a new multimedia retrieval system which integrates local diversity into an evolving global knowledge. A suitable personalization of global ontologies allows to extract information from multimedia content according to context-sensitive relevance, while a peer-to-peer communication model provides a decentralized and scalable logical architecture for media search.