Injecting KGs in RAG in pre-processing, post-processing, chunk extraction, document and contextual hierarchies, with a concrete example of how all these techniques can produce more accurate results.