SemFlowRAG: Directed Semantic Flow from Abstraction to Evidence for Complex Reasoning

2026-06-26

AI Summary

This work addresses the “probability sink” problem prevalent in existing knowledge graph–based retrieval-augmented generation methods, which often leads to semantic drift and noise accumulation during multi-hop reasoning. To mitigate this issue, the authors propose a Semantic Gradient Graph framework that quantifies the semantic abstractness of entities to construct a directed, hierarchical retrieval structure. They further introduce an abstractness-guided, directed PageRank algorithm to enable a targeted semantic flow from high-level abstract concepts down to low-level evidential facts. By integrating knowledge graphs, embedding variance metrics, and a data-driven graph construction strategy, the proposed approach significantly alleviates the probability sink phenomenon. Experimental results on complex question answering benchmarks demonstrate consistent improvements over current baselines in both retrieval accuracy and downstream reasoning performance.

Abstract

Retrieval-Augmented Generation (RAG) enhanced by Knowledge Graphs has shown promise in complex multi-hop reasoning tasks. However, existing graph-based retrieval methods typically rely on flat, undirected topologies. During the retrieval process, the probability flow often gets trapped in high-degree abstract concept nodes which we define as ``probability black holes'', leading to semantic drift and noise accumulation. To address this, we propose SemFlowRAG, a framework that reconstructs the flat retrieval space into a corpus-adaptive semantic gradient graph. This data-driven self-organization enables a hierarchical structure to emerge naturally from the data distribution, capturing the intrinsic semantic granularity of the corpus to suppress structural noise. By quantifying the semantic abstractness of entities through the embedding variance of their associated passages, we transform static undirected edges into directed semantic constraints. Furthermore, we design an abstractness-guided directed PageRank algorithm that forces the retrieval trajectory to follow a ``high-to-low semantic abstractness'' gradient. This mechanism ensures layer-by-layer evidence convergence, smoothly guiding the retrieval process from abstract concepts to specific document evidence. Extensive experiments on complex QA datasets demonstrate that SemFlowRAG effectively mitigates the ``probability black holes'' issue, outperforming existing baselines in both retrieval and downstream reasoning performance.

Problem

• Retrieval-Augmented Generation
• Knowledge Graphs
• multi-hop reasoning
• semantic drift
• probability black holes

Innovation

• Semantic Gradient Graph
• Directed PageRank
• Probability Black Holes
• Abstraction-guided Retrieval
• Corpus-adaptive Self-organization