I'm experimenting with N3.js (RDF.js) and I see that it provides an API for storing triples in memory and accessing/updating/working with them. what is a good way to instead work with triples stored in a database?
I see that N3.js also provides a means of serializing/deserializing between in-memory triples and RDF documents. I also see that there are a few DBMSs that seem to be specifically for working with persisted triple storage (GraphDB, and presumably others).
Is the intention behind the N3.js project that users work with the file system directly?
what is a good way to instead work with triples stored in a database?
There are lots of RDF databases, "triplestores", to choose from, Jena, GraphDB that you mention, AllegroGraph, AWS Neptune, RDF4J... These all store RDF natively and RDF is always RDF (i.e. perfectly interchangeable) so you can use any triplestore and change your mind to use another and easily move the data over.
Is the intention behind the N3.js project that users work with the file system directly?
Not necessarily: it's a parsing tool for certain forms of RDF data but where you put it is up to you!