Validator & optimizer

Greetings,

I am familiar with the schema validator at: https://validator.schema.org

I am wondering if there has been any work (theoretical or
practical/functional) on creating a tool which optimizes JSON+LD graphs.
In the case of schema.org and the use of the vocabulary it is possible to
list assertions several times in the graph. Alternatively, one could
cross-reference these entities. For example, maybe a person is listed with
a description of who they are and some basic facts and relationships, this
person could be listed as a director of a film listed on the same page
along with their details, or that part of the graph could just reference
the person in the graph somewhere else.

Obviously the cross-listing/referencing strategy leads to an equally rich,
but less verbose graph. I'm looking for a tool which would catch these
sorts of "double listings"... does something like that exist?

all the best,
- Hugh Paterson III