Skip to Main Content
Enabling the "Web of Data" has recently gained increased attention, particularly driven by the success of Linked Data. The agreed need for technologies from the database domain is therein often referred to as the "Web as a Database", a concept that is still more a vision than a reality. Meanwhile, the database community proposed the notion of dataspaces managed by support platforms, as an alternative view on the data management problem for small-scale, loosely connected environments of heterogenous data sources. The Web of Data can actually be seen as a collection of inter-connected dataspaces. In this work, we propose a combination of Linked Data and database technologies to provide support platforms for these Web dataspaces. We argue that while separated, Linked Data still lacks database technology and the dataspace idea lacks openness and scale. We put particular focus on the challenge of how to index, search and query structured data on the Web in a way that is appropriate for its dynamic, heterogeneous, loosely connected, and open character. Based on an empirical study, we argue that none of the two extremes on its own - centralised repositories vs. on-demand distributed querying - can meet all requirements. We propose and discuss an alternative hybrid approach combining the best of both sides to find a better tradeoff between result freshness and fast query response times.