Enhancing Job Matching With Knowledge Graphs and RAG
summary
Builds a Neo4j knowledge graph of profiles, roles, skills, companies and recruitment journeys and demonstrates two concrete RAG pipelines: one using ChromaDB vector strategy to map natural‑language questions to Cypher via example pairs (author reports >100 examples markedly reduce hallucination), and another storing embeddings inside Neo4j so the DB itself returns k nearest profiles/roles for prompting an LLM. Shows generated Cypher that needed iterative fixes, explains hybrid use to handle mismatched strings, and calls out 1–2s API latency as a scaling issue.
AI-written summary · about
chapters (9)
- 0:00Introduction to job matching
- 1:39Enhancing systems with RAG
- 3:34Knowledge graphs and Neo4j
- 6:05Retrieval-augmented generation
- 8:38Defining the graph domain
- 10:14Cypher generation strategy
- 17:15Embedding nodes in Neo4j
- 21:34Combining strategies
- 23:04Challenges and future work
related talks
24mGraph-Powered Job Matcher: From SQL Tables to Neo4j RAG in Production
27mAdaptive Knowledge Graphs for Dynamic RAG Systems
39mEnhancing RAG with Neo4j Knowledge Graph
53mPractical GraphRAG - Making LLMs smarter with Knowledge Graphs
30mAgentic RAG with Knowledge Graphs Explained
26mPractical GraphRAG Making LLMs smarter with Knowledge Graphs
01h 40mKnowledge Graphs & GraphRAG: Techniques for Building Effective GenAI Applications
16mEnhancing Retrieval-Augmented Generation with GraphRAG Patterns in Neo4j