![[2510.20505] RELOOP: Recursive Retrieval with Multi-Hop Reasoner and Planners for Heterogeneous QA](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2510.20505] RELOOP: Recursive Retrieval with Multi-Hop Reasoner and Planners for Heterogeneous QA
Summary
The paper presents RELOOP, a novel framework for recursive retrieval in heterogeneous question answering (QA) that enhances efficiency and accuracy through structured iteration and evidence synthesis.
Why It Matters
RELOOP addresses the challenges of retrieval-augmented generation (RAG) in multi-step questions and diverse data formats. By improving the efficiency of evidence collection and synthesis, it has the potential to significantly enhance the performance of QA systems across various domains.
Key Takeaways
- RELOOP utilizes a hierarchical sequence approach for structured data retrieval.
- It improves QA accuracy while reducing unnecessary computational overhead.
- The framework is format-agnostic, allowing it to work across text, tables, and knowledge graphs.
- Guided iteration helps in budget management for tool calls and token usage.
- Evidence canonicalization enhances answer consistency and auditability.
Computer Science > Computation and Language arXiv:2510.20505 (cs) [Submitted on 23 Oct 2025 (v1), last revised 26 Feb 2026 (this version, v3)] Title:RELOOP: Recursive Retrieval with Multi-Hop Reasoner and Planners for Heterogeneous QA Authors:Ruiyi Yang, Hao Xue, Imran Razzak, Hakim Hacid, Flora D. Salim View a PDF of the paper titled RELOOP: Recursive Retrieval with Multi-Hop Reasoner and Planners for Heterogeneous QA, by Ruiyi Yang and 4 other authors View PDF HTML (experimental) Abstract:Retrieval-augmented generation (RAG) remains brittle on multi-step questions and heterogeneous evidence sources, trading accuracy against latency and token/tool budgets. This paper introduces RELOOP, a structure aware framework using Hierarchical Sequence (HSEQ) that (i) linearize documents, tables, and knowledge graphs into a reversible hierarchical sequence with lightweight structural tags, and (ii) perform structure-aware iteration to collect just-enough evidence before answer synthesis. A Head Agent provides guidance that leads retrieval, while an Iteration Agent selects and expands HSeq via structure-respecting actions (e.g., parent/child hops, table row/column neighbors, KG relations); Finally the head agent composes canonicalized evidence to genearte the final answer, with an optional refinement loop to resolve detected contradictions. Experiments on HotpotQA (text), HybridQA/TAT-QA (table+text), and MetaQA (KG) show consistent EM/F1 gains over strong single-pass, multi-hop, and ...