The World Wide Web provides an enormous amount of images easily accessible to everybody. The main challenge is to provide efficient search mechanisms for image content that are truly scalable and can support full coverage of web contents. In this paper, we present an architecture that adopts the peer-to-peer (P2P) paradigm for indexing, searching and ranking of image content. The ultimate goal of our architecture is to provide an adaptive search mechanism for image content, enhanced with learning, relying on image features, user-defined annotations and user feedback. Thus, we present PIRES, a scalable decentralized and distributed infrastructure for building a search engine for image content capitalizing on P2P technology. In the following, we first present the core scientific and technological objectives of PIRES, and then we present some preliminary experimental results of our prototype.
展开▼
