It is a bit surprising to see PDF/OOXML parser, a graph database, etc crammed into an inference engine, why do you need to put them together?
As we write in the docs, they are purely there to support the Studio and not the pure inference engine. Studio runs in as separate binary and helps users to easily test advanced features of models.