Linking actual data with dataset catalogue metadata

[letmaik]

One of the issues with dataset catalogue metadata (#6) is that there is some friction to link actual data to them such that it can also be used in some way by end-users, web applications etc. This discussion should identify these issues and suggest solutions. (subdiscussion of #3)

I'll start with the case of a LOD-only dataset. In that case, if the VoID metadata ontology is used, it allows to point to a SPARQL endpoint, and this could be where the actual data is. It is very generic though and may not fit to everything. The client must follow this link then and see if it can make sense of the data it finds in the SPARQL endpoint. Another question would be: Is every client supposed to understand (Geo)SPARQL or can data also be exposed somehow else?

Another issue is, do we deal with actual downloadable files at some point? For some WP, probably, so these files have to be described and their URL possibly put into LOD in some organised way as well. Are there any experiences here?

[jonblower]

I think we should publish GeoSPARQL when we can, but that will probably only be feasible for the vector datasets. For others (e.g. raster), other APIs or file-download endpoints may be needed.