Rdfsync: Efficient Remote Synchronization Of Rdf Models

In this paper we describe RDFSync, a methodology for efficient synchronization and merging of RDF models. RDFSync is based on decomposing a model into Minimum Self-Contained graphs (MSGs). After illustrating theory and deriving properties of MSGs, we show how a RDF model can be represented by a list of hashes of such information fragments. The synchronization procedure here described is based on the evaluation and remote comparison of these ordered lists. Experimental results show that the algorithm provides very significant savings on network traffic compared to the file-oriented synchronization of serialized RDF graphs. Finally, we provide the design and report the implementation of a protocol for executing the RDFSync algorithm over HTTP.