Convert a document corpus into a normalized semantic embedding index optimized for similarity-based retrieval. Accepts t