It's one of the problems in the RDF/SPARQL world that you don't really want to update the database a triple at a time (even guarded by transactions) but you want some concept of a record.
To say that one wants some concept of a record is to say that one wants some concept of a process. But the relational paradigm is understood and taught in terms of overwhelming incentives to overload the types that represent process steps, therefore conventional data modelling makes process integrity effectively impossible.
