Peer-to-peer data integration using distributed bridges.

摘要

The traditional methods of data integration use a centrally administered global schema that is not well-suited to a large number of data sources that frequently change. Using the peer-to-peer approach, this thesis proposes the use of user-contributed mappings or 'bridges' that can be published and exchanged over a network of peers. The Universal Peer-to-Peer (U-P2P) framework is modified to allow bridges to be created and a schema definition for the bridges is presented that contains mappings and a semantic relationship. Large databases are connected to the peer network using proxy modules that allow U-P2P to connect to a live data source using a custom implementation. A fully decentralized network adapter is implemented in U-P2P to prevent large databases from having to upload their metadata to indexing servers. The proposed system is used to implement simple data integration between two digital repositories using query translation.