Allow me to try rephrasing this:

I'm searching for a strong RDF store or library using the following features:

  • Named graphs, as well as other type of reification.
  • Version monitoring (most likely in the named graph level).
  • Privacy between categories of customers, either at named graph or triple level.
  • Human-readable data input and output, e.g. TriG parser and serialiser.

I have performed with Jena, Sesame, Boca, RDFLib, Redland and a couple of others a while ago but each had its problems. Have enhanced within the above areas lately? Can other things do things i want, or perhaps is RDF not ready for prime-time?

Reading through round the subject a little more, I have discovered that:

Talis may be the apparent choice, but privacy might be an problem, or perceived problem anyway, since its a SaaS offering. I only say apparent since the three emboldened features inside your list are core options that come with their platform IIRC.

They do not possess a features list as a result - which causes it to be tough to support this answer, but they are doing say that stores of information could be individually guaranteed. I guess you can - in a pinch - join another store with respect to each of your customers.

Human readable input is frequently best based on writing custom connects for every user-task, which means you best be ready to do this as needs demand.

Regarding prime-time readiness. I'd say yes for many programs but otherwise "less than". Mostly the city must integrate with existing developer toolsets and write good documentation targeted at "regular" designers - most likely OO designers using Java, .Internet and Ruby/Groovy - after which I predict it'll snowball.

See also Temporal Scope for RDF triples