Directions and Challenges for Semdata

The data driven world is here. What will RDF’s place therein be? In this paper, we explore the inevitable coming together of RDF, analytic oriented databasing, web scale computing and reasoning. To warehouse, analyze, publish and subscribe to anything anywhere, draw conclusions and distill knowledge. This is the challenge. As a first part of a response, we show how we attain a space consumption of just under 6 bytes per quad of RDF data extracted from the archetypal TPC H benchmark data set and how we get comparable query times with the data stored in relational and RDF forms in Virtuoso. This is achieved with a new column-wise compressed storage subsystem to be at the core of later Virtuoso releases. This represents a more than fourfold improvement over previous RDF space efficiency. While RDF scalability in absolute terms has been resolved with the advent of many cluster capable RDF stores, the advances outlined here radically reduce the cost of RDF when compared with its technological alternatives. We then look at ways of synchronizing RDF data sets across wide area distances and at ways of keeping RDF extractions of local relational data up to date. We conclude with a preview of research to be undertaken in LOD2, a EU FP7 project starting in the fall of 2010. Among other topics, this will address RDF exploitation of many relational techniques pioneered at in MonetDB at CWI, one of the project partners. On the basis of all this, we are confident that RDF will strengthen its role in the coming data economy.