Knowledge GraphsRetrievalQuestion AnsweringLangChainRDF

Knowledge-graph and hybrid retrieval for course enrollment QA

Built a question-answering system for course enrollment using RDF and SPARQL alongside LangChain and FAISS for hybrid retrieval.

2024-11-011 min read

Overview

Answer course-enrollment questions with a system that can combine structured academic rules and semantic search.

System design

  • Modeled course, prerequisite, and program information as RDF triples
  • Queried structured relationships with SPARQL for deterministic factual lookup
  • Added LangChain and FAISS for semantic retrieval over course descriptions and related text
  • Combined graph-based and vector-based retrieval into a hybrid answering flow

Results

  • Compared graph-only, vector-only, and hybrid retrieval behavior on enrollment questions
  • Improved coverage for questions that mixed strict prerequisites with natural-language intent
  • Built a reusable hybrid retrieval pattern for domain-specific QA tasks