PAGE: A Distributed Infrastructure for Fostering RDF-Based Interoperability

This paper shows how to build a scalable, robust and efficient distributed Internet-scale RDF repository, that we name PAGE (Put And Get Everywhere). 1 Motivation In the recent years, the RDF (Resource Description Framework) data model is gaining momentum. Among other significant examples, take for istance the adoption of key players such as Adobe (i.e., the eXtensible Metadata Platform XMP) and Oracle (i.e., the 10g version of the famous RDBMS promise to be the best RDF storage engine). RDF enables the encoding of relational data over the Internet using triples usually denoted with (spo), where s is the subject, p is the predicate and o the object. Complex structures can be easily encoded in a set of RDF triples. But, what about managing them? Centralized solutions may help in the short term, but if RDF should become the basis of the Semantic Web, some sort of distribution must be take into account. What if we could put and get triples without bothering of the underlying infrastructure because such infrastructure is the whole Internet? Such a distributed and decentralized RDF repository is not yet available, but the peer-to-peer community has been studying the “turn on, put() and get() out” paradigm using Distributed Hash Tables (DHTs) for almost a decade. In this paper we describe PAGE (Put And Get Everywhere), an optimized peerto-peer infrastructure for distributed RDF storing and retrieval: an original approach that comes out from the convergence between the current works on DHT in the peer-to-peer community and on optimized index structures for querying RDF in the Semantic Web community named YARS [1]. Our work aims at advancing the state-of-the-art in decentralized RDF repositories finding a convergence between peer-to-peer (i.e., DHT) and Semantic Web researches. In particular it belongs to the class of solutions which avoid flooding approach, previously investigated by RDFPeers [2] and GridVine [3].