[Submitted on 28 May 2026 (v1), last revised 15 Oct 2026 (this version, v3)]
Abstract:Retrieval over semi-structured knowledge bases (SKBs) must handle both textual descriptions and relational constraints. Text-based retrievers capture the overall semantics of a query but struggle with multi-hop relations, whereas graph-based retrievers follow relational paths precisely but miss the broader meaning. We propose GRASP, a hybrid retriever that learns to fuse the two. A small LLM planner translates the query into an executable graph plan in a single call, its graph ranking is fused with a text ranking, and a fine-tuned reranker reorders the fused candidates. Instead of training the planner to be accurate on its own, we optimize it with reinforcement learning to maximize the gain its graph ranking brings to the fused ranking. On three STaRK benchmarks, GRASP achieves the best results on all metrics, improving the average Hit@1 by 18.6 points over the strongest baseline, while taking only 2.47 seconds per query.
Submission history
From: Yicheng Tao [view email]
[v1] Thu, 28 May 2026 17:07:32 UTC (7,041 KB)
[v2] Sun, 31 May 2026 00:43:51 UTC (7,041 KB)
[v3] Thu, 15 Oct 2026 17:59:16 UTC (966 KB)
