i.e., if a media item X references licence URI Y which hasn't yet been ingested, we should store the fact that data relating to X will be impacted when Y is eventually encountered. If we do this at the point that X is updated, we can make this a great deal more efficient than SPARQL queries later on.
i.e., if a media item X references licence URI Y which hasn't yet been ingested, we should store the fact that data relating to X will be impacted when Y is eventually encountered. If we do this at the point that X is updated, we can make this a great deal more efficient than SPARQL queries later on.