Friday, December 10, 2004

Provenance to Unify Browsing

A Framework for Unified Information Browsing via RDFX "This Unified Information Service Browser (UISB) project is investigating the means of joining heterogeneous information from multiple sources into a combined and navigable structure...Attempting to unify disparate data sources illustrates that hierarchical and relational data models can place undesirable constraints on the stored data; however, the same data can be stored without constraint in a dynamic graph, which also provides significant potential for enhanced traversal and discovery capabilities."

"The solution that was adopted was to include provenance information for every statement that is added to the graph. Thereafter if a graph, or a statement or a resource is removed from the store, it can be done without the risk of losing information. In addition to identifying where a statement originated, the provenance objects can store metadata describing (among other things) when it was discovered, and when it should expire, therefore enabling the automated management of expired information, or potentially providing an event mechanism to instigate the refresh of the data. The benefit of this solution, is that it does not rely on the graph implementation to ensure that graph merging and un-merging happens as expected, however the added overhead of four extra statements for tracking provenance may negatively affect performance and increase the size of the data store. This will depend on the implementation of the graph store since some stores (e.g. Jena 6.4) can reify statements more efficiently."

No comments: