SubQRAG: Sub-Question Driven Dynamic Graph Rag

Graph Retrieval-Augmented Generation (Graph RAG) is effective at building a knowledge graph (KG) to connect disparate facts across a wide document corpus. However, this broad-view approach often lacks the deep structured reasoning required for complex multi-hop question answering (QA), leading to incomplete evidence and error accumulation. To address these issues, we propose SubQRAG1, a sub-question driven framework that enhances reasoning depth. SubQRAG decomposes a complex question into an ordered chain of verifiable sub-questions. For each sub-question, it retrieves relevant triples from the graph. A key feature is its ability to adapt: if the original graph is insufficient, the system retrieves information from source documents, extracts new triples, and dynamically updates the graph in real time. Finally, all triples used in the reasoning process are aggregated into a "graph memory", providing a structured and traceable evidence path for the final answer generation. Experiments on three multi-hop QA benchmarks show that SubQRAG consistently achieves significant improvements, particularly in Exact Match scores.2