Misconception Mapper RAG is an

Misconception Mapper RAG is an interpretable education-focused RAG system for working with standards, rubrics, readings, assignments, and anonymized student work. It pairs a Next.js interface with a FastAPI backend and a hybrid retrieval stack combining dense vector search, BM25, Reciprocal Rank Fusion, and CrossEncoder reranking.

Unlike a generic document chatbot, the application produces purpose-built student and teacher outputs while keeping its evidence trail visible through citations, source metadata, confidence levels, missing-evidence flags, and an end-to-end retrieval debugger.