Options
2003
Conference Paper
Title
Efficient data store discovery in a scientific P2P network
Abstract
In this paper we propose to use schema-based peer-to-peer systems for the exchange of scienitific documents. Compared with recent file sharing networks highly domain-oriented scientific documents are shared in relatively small research communities with tight internal and only a few external connections. The documents are annotated with small a but welldefined sets of metadata using standard taxonomies and are stored in many distributed, autonomous, heterogenous data repositories. As a recent development super-peer networks emerge improving the network performance by clustering peers at super-peers. By using Semantic Overlay Clusters (SOC) for clustering the super-peer networks we enable the creation of context-specific, logical views over the physical P2P network topology according to the research communities demands. SOCs define peer clusters according to the metadata description of peers and their contents. The respective clustering policy expresses the demand on the peers for a particular research context. In this paper we show how SOCs can be defined based on policies. We detail the definition of conditions that we use for our ECA-rule like policiy definition approach.