I have a software that generates an RDF representation of certain dataset. I want to add to the generated data also some metadata describing not specific data contained in the data set but the document itself - i.e., when the document was created, by which software, which version, etc. The schema.org properties provide the necessary relationships, but I can not figure out the proper place to attach it. Is there some standard way of saying "this is the metadata about the document itself" in RDF? I use Turtle serialization for RDF but generic answer working with any serialization would be preferable.
There is not a standard place to do this. A RDF graph is just a collection of triples; it's not identified by an IRI or anything like that. (However, in SPARQL datasets, you could post some metadata about a named graph by using the name of the graph as the subject in a triple. That would just be a convention, though. It's not "official" in any sense.)
In the RDF serializations of OWL ontologies, there can be an ontology element (i.e., a resource with the type owl:Ontology), and that can be used to associate some metadata with the ontology. You'd probably want to adopt an approach like that. That is, you'd establish a convention with something like
@prefix ex: <...>
[] a ex:DatasetRepresentation ;
ex:created "..." ;
ex:representationOf <...> .
#... rest of generated content ...