Most Open Government Data initiatives are centralised and unidirectional (i.e., they release data dumps in CSV or PDF format). Hence for non trivial applications reusers make copies of the government datasets, to curate their local data copy. This situation is not optimal, leads to duplication of efforts and reduces the possibility of sharing improvements. To improve the usefulness of publishing open data, several authors recommeded to use standard formats and data versioning. Here we focus on publishing versioned open linked data (i.e., in RDF format) because they allow one party to annotate data released independently by another party thus reducing the need to duplicate entire datasets. After describing a pipeline to open up legacy-databases data in RDF format, we argue that RDF is suitable to implement a scalable feedback channel, and we investigate what steps are needed to implement a distributed RDF-versioning system in production.

Link to proceedings (the paper is at page 171):