Building a Distributed Infrastructure for Scalable Triple Stores

Built specifically for the Semantic Web, triple stores are required to accommodate a large number of RDF triples and remain primarily centralized. As triple stores grow and evolve with time, there is a demanding need for scalable techniques to remove resource and performance bottlenecks in such systems. To this end, we propose a fully decentralized peer-to-peer architecture for large scale triple stores in which triples are maintained by individual stakeholders, and a semantics-directed search protocol, mediated by topology reorganization, for locating triples of interest. We test our design through simulations and the results show anticipated improvements over existing techniques for distributed triple stores. In addition to engineering future large scale triple stores, our work will in particular benefit the federation of stand-alone triple stores of today to achieve desired scalability.