GRASP: Plan-Guided Graph Retrieval with Adaptive Fusion and Reranking on Semi-Structured Knowledge Bases

2026-05-28

AI Summary

Existing approaches to semi-structured knowledge base retrieval struggle to effectively integrate graph structure with textual information, limiting the precision of path-based retrieval. This work proposes GRASP, a novel framework that deeply integrates query planning into the graph retrieval pipeline for the first time. GRASP achieves end-to-end optimization through plan-guided graph traversal, an adaptive fusion mechanism conditioned on the query plan, and a fine-tuned reranker. Evaluated on the STaRK benchmark, the method sets new state-of-the-art results across all three tasks, improving the average Hit@1 from 62.0 to 73.9 and substantially enhancing both retrieval accuracy and robustness.

Abstract

Semi-structured knowledge bases (SKBs) embed textual documents in a typed graph of entities and relations, and underpin applications such as product search, academic paper search, and precision-medicine inquiries. Existing hybrid retrieval systems on SKBs either use the graph only for query expansion, mix textual and structural branches under a global weighting, or rely on fine-tuned graph-traversal generators. We present GRASP, a three-stage SKB retrieval framework unifying plan-based graph retrieval, plan-conditioned fusion with a dense retriever, and a fine-tuned reranker over the fused candidates. GRASP substantially advances the state of the art on every metric across the three STaRK benchmarks, lifting average Hit@1 from 62.0 to 73.9. Ablation and sensitivity studies further confirm the effectiveness and robustness of GRASP.

Problem

• semi-structured knowledge bases
• retrieval
• graph retrieval
• hybrid retrieval
• information retrieval

Innovation

• plan-guided retrieval
• adaptive fusion
• semi-structured knowledge base
• graph-text retrieval
• reranking