RDF to RDB (and back)
So, everybody who has worked with linked data knows about conversions from and to other formats. From relational databases (RDB) or flat files, etc., to RDF, because, well, that is the format we need for networked data. And the other way around as well, from RDF to RDB, for example, because data visualization tooling works on relational data, not linked data.
The main problem is that when we do conversions like that, we actually copy data, be it in a different format. But the data we originally only had in one system, we will have after the conversion in two systems as two separate sets. This creates various problems:
- If the data is big, we’ll have two big sets.
- But more importantly, we’ll have two sets we’ll have to keep up to date.
The last point is usually the most difficult one, especially when the data is in an operational system, where it is changed by data transactions by users continuously. Usually, the only feasible way is to copy data periodically from one format to the other, which creates another complexity. We need to keep in mind the date and time of the moment of the conversion and must remember that the most recent data is not in both systems.
Data at the source
In the Netherlands, the country where I live, there is a strong architectural direction, called “data at the source” (in Dutch: “data bij de bron”). The main idea is to avoid copies of data. The line of thought here is that having no copies solves a lot of data handling problems, among them the problems we indicated above.
Here in the pages of this blog, we’ll investigate options to use RDF data without the dreaded copying.
We’ll look at two options: