HDT Quads: Compressed Triple Store for Linked Data
Details of idea
Teilnahme am Wettbewerb Wissenschaftspreis - Universitäten/Fachho ... (Öffentlich)
Beschreibung:
The Linked Data philosophy extends the Web through standards promoting semi-structured data publication, exchange and consumption. The main standard for modeling data on the Web is RDF, a graph-based model for relationships between resources. With increasing size, compressibility and queryability of RDF graphs become an issue. This is addressed by the HDT (Header-Dictionary-Triples) format, a compressed representation of RDF datasets supporting retrieval features without decompression. Yet, data often contains additional graph information, such as the origin or version of a triple, which traditional HDT cannot handle. This work introduces HDTQ (HDT Quads), a highly compact and queryable extension of HDT, which is able to represent additional graph information. Two approaches of HDTQ, Annotated Triples and Annotated Graphs, are introduced and the performance is compared with leading open-source RDF stores, Apache Jena and Virtuoso. Results show HDTQ as a highly competitive alternative.