Compact Representation of Large RDF Data Sets for Publishing and Exchange

Increasingly huge RDF data sets are being published on the Web. Currently, they use different syntaxes of RDF, contain high levels of redundancy and have a plain indivisible structure. All this leads to fuzzy publications, inefficient management, complex processing and lack of scalability. This paper presents a novel RDF representation (HDT) which takes advantage of the structural properties of RDF graphs for splitting and representing, efficiently, three components of RDF data: Header, Dictionary and Triples structure. On-demand management operations can be implemented on top of HDT representation. Experiments show that data sets can be compacted in HDT by more than fifteen times the current naive representation, improving parsing and processing while keeping a consistent publication scheme. For exchanging, specific compression techniques over HDT improve current compression solutions.