SelfGraphRAG: Bridging the Supervision Gap in Graph-Based RAG with Synthetic QA Generation

Researchers introduced SelfGraphRAG, a framework that generates synthetic question-answer pairs from knowledge graph structure to train a query-conditioned graph retriever, addressing the lack of labeled data for graph-based RAG. Experiments on multi-hop QA and classification benchmarks showed improved retrieval precision and downstream reasoning over embedding-based baselines, indicating that graph structure can provide supervision when labeled data are unavailable.

arXiv:2608.25123v1 Announce Type: new Abstract: Retrieval-augmented generation RAG improves large language models by incorporating external knowledge without retraining, but existing methods often underuse the relational structure encoded in knowledge graphs. Graph-based RAG can capture entity relationships, yet supervised graph retrieval typically requires labeled question-answer data that may not be available for newly constructed graphs. We address this limitation with SelfGraphRAG, a framework that generates question-answer pairs directly from knowledge graph structure and uses them to train a query-conditioned graph retriever. The generated questions capture multi-hop paths and local neighborhoods, providing relational supervision without manual annotation. Experiments on multi-hop question answering and classification benchmarks show that SelfGraphRAG improves retrieval precision and downstream reasoning performance over embedding-based baselines. These results suggest that knowledge graph structure can provide useful supervision for training graph retrievers when labeled data are unavailable.