FlowRAG: Synergizing Explicit Reasoning via Frequency-Aware Multi-Granularity Graph Flow

FlowRAG is a newly proposed framework that enhances graph-based retrieval-augmented generation (GraphRAG) by integrating a quad-level heterogeneous graph structure, which includes nodes for passages, summaries, sentences, and entities. It features a dual-granularity activation module for improved semantic recall and a frequency-aware weighted flow module that optimizes relevance routing through entity-passage links, yielding robust multi-hop reasoning capabilities. This approach demonstrates state-of-the-art performance on complex reasoning benchmarks, making it a significant advancement for practitioners focusing on knowledge-intensive AI tasks.