A. Löser (Germany)
: Distributed E-Learning Repositories, Peerto-Peer Data Management Systems, Architectures forWeb-based Learning Systems
Peer-to-Peer data management approaches perform well as frameworks for large networks of autonomous, heteroge nous and global distributed data sources. However, for a given query, limited system resources prohibit forwarding a query to all possible peers. The key to scalability is to send queries only to relevant data sources thus avoiding over flodding the network with queries. The key to scalability is to send queries only to relevant data sources thus avoiding overflodding the network with queries. Due to the users's lack of global knowledge which sources are relevant and the unpredictable dynamics of available data sources users should automatically notified when a new data source con nects to the network. In this paper we investigates adaptive semantic overlay views, user specific overlay structures of relevant data sources. We discuss what models are suitable for describing information sources as well as user profiles, how the models and user profiles are stored, queried and managed in a distributed catalog and how to adaptively compute the overlay structure. Due to the tight integration of a symmetric peer-to-peer based publish/subscribe ar chitecture with fast matching techniques for content-based Metadata, our architecture is capable to manage a large number of data sources and users.
Important Links:
Go Back
